Posted by Larry Hoover on May 18, 2009, at 11:25:41
In reply to Re: Here's what I think., posted by alexandra_k on May 18, 2009, at 7:00:31
> There is a way of designing a trial such that two outcomes are possible: The first outcome is that there is support for the efficacy of the medication (compared to placebo and alternative medication). The second outcome is that the study failed to find support for the efficacy of the medication (compared to placebo and alternative medication). The crucial thing about this is supposed to be that the second outcome DOESN'T entail or imply that the medication that was trialed is WORSE or MORE HARMFUL than either placebo or alternative medication and it also DOESN'T entail or imply that further studies (maybe 5 in 100) will find the first outcome - that there is support for the efficacy of the medication.
The simplest methodology is drug against placebo. The underlying assumption is that there is no difference, called the null hypothesis. The methodology is designed to standardize everything they think might be relevant, such that going into the study, both groups are identical as can be. If the results of the study indicate that the two groups differ significantly after the treatments, then the null hypothesis is rejected. If that is the case, it is generally hoped that the drug was found to be superior, rather than the other way around.
The methodology you describe has three arms, investigational drug, placebo, and active comparator. In such a trial, any one arm could be found to be significantly different from one or both of the other arms. There are three null hypotheses which may be rejected.
It is quite correct that acceptance of the null hypothesis, i.e. no significant difference found, is not evidence that there is no difference. It is an aphorism in scientific research, "Absence of evidence is not evidence of absence."
Now, with respect to the repetitious testing of the null hypothesis after non-significant results, there is a financial penalty for continuing with further trials. If the first trial(s) showed a strong trend toward efficacy, then further trials, usually with more subjects, might be warranted. But 90-95% of investigational psych drugs simply don't pan out, more often at Phase 1 (safety), than at Phase 2 (dose-ranging), or Phase 3 (efficacy).
> IF this is correct (that there are ways of designing trials such that you either show efficacy or you fail to find efficacy) THEN is it the case that researchers need to design their trials in this fashion in order to obtain pharma funding (or free / subsidized samples of expensive medications) to do the trial? If so, then this would be pharma driving scientific methodology in a way that obviously benefits them (they are required to report HARMFUL findings but they are not required to report FAILURE TO FIND POSITIVE BENEFITS). This is of course important because if you run a study often enough, well, 5 in 100 will give you the result you seek).
I don't know of any other way to do it, other than I described. No company is going to fund 100 studies to get 5 good ones. All research is now registered and in the public domain, so there can be no hiding of negative outcomes.
> Deaths would be something that you would think would be hard to have not show the medication to be positively harmful. But the same studies that showed certain anti-depressants to be more effective than alternative and placebo were the studies that (pharma didn't think) showed certain anti-depressants to result in more deaths in certain age groups.Where are you getting the idea that there were deaths? For example, there were no deaths ever recorded in adolescent antidepressant trials.
I well recall when the lay press first reported some statistic that adolescent trials of Paxil had demonstrated a seven-fold increase in suicidality (not completed suicide). I dug up the actual clinical trial data, some 500 or 600 pages of records (if I recall correctly), rather than the 6 or 7 pages in the published report. On examining the raw data, the reported increase in suicidality was absolutely false.
The methodology for the study in question had detailed requirements for the reporting of side effects and misadventure. The category which included suicidality was called "extreme emotional lability." But it also included other violations of study protocol. I clearly recall three such reports which you might find revealing.
In the drug group, one such report was of a young lady who took more of the drug than she should have. For some strange reason, each participant was supplied with 20 foil-pack capsules every two weeks, even though they were to take one capsule per day. They were to turn in the remainder when picking up the next batch. One young lady turned in less than the expected amount of unused capsules, and she was warned to follow the protocol. Four weeks later, it happened again. Because she was warned, it was coded as "extreme emotional lability", according to the a priori reporting guidelines. The individual improved significantly on the medication, and went on to complete the continuation phase of the study.
Another female randomized to the drug arm lied about numerous prior para-suicidal behaviours (an exclusionary factor) on the study application. She stopped taking the medication on her third day, but did not report that to the study investigators, which would have led to her being recorded as withdrawn. Three weeks later, she had a fight with her mother, and swallowed a small overdose of Tylenol (her preferred para-suicidal behaviour). Even though she had lied to get into the study, and lied about taking the study medication, because she had not formally withdrawn, her behaviour was recorded as a suicidal attempt in the medication group.
The only real suicidal act, IMHO, was recorded in the placebo group. Another female participant had failed to respond to her treatment, and had slashed her wrists. Upon being admitted to hospital, her blind was broken so as to provide her with appropriate care. She was administered an antidepressant (not the study med), and recovered thereafter.
I just don't recall all the other incidents that were lumped into this category, but in my interpretation, the score should have read 0 reports of suicidality in the medication arm, and 1 in the placebo arm. Or, at most 1 and 1, as the liar was a participant that might have been randomized to either group. By a flip of the coin, it could have been 0 and 2. In any case, I found no evidence of the reported "seven-fold increase in suicidality" trumpeted in the lay press.
> Was it a genuine oversight to fail to pick up on this in the discussion section or would discussing such a thing jepordize ones future research funding?
I have never known such outcomes to not be reported, but I have not read every similar study.
> I thought (could well be wrong) that the 'gold standard' of treatment for psychiatry is the last generation. So... If we want to get a new 'erectile dysfunction' medication approved then the way to go about it would be to either show it to be more effective than viagra OR show that it has less harmful side-effects (not quite sure how the latter works out - but I heard the latter was the main reason for the move from MAOI's to SSRI's, major tranquilisers to new gen anti-psychotics, and lithium to alternative mood stabilizers. That efficacy really wasn't any better, but the side-effects were less? Not sure how compatible this is with what I said before).
The active comparator and placebo methodology is generally preferred because if the active comparator also fails to show superiority over placebo, the study methodology needs a much closer look. Phase 3 is efficacy, so that is the primary consideration. If reported side effects are also lesser, then that is a bonus for marketing thereafter. If reported side effects are higher, then approval might be withheld, nothwithstanding efficacy.
The move towards SSRIs was generally due to lower toxicity vis a vis tricyclics, with the lower side effect profile being an added feature.
> Yes, one needs to run a profitable business, I understand that. Yes, it costs a lot of money to develop new drugs (and there need to be lots and lots of trials when one is trying to get a good result from a number of trials from a drug that simply isn't terribly effective). Yes, it costs a lot of money to advertise the products (millions if not billions). It is kinda hard to measure 'profit' but really... Are they just making 'profits' or are they basically 'profiteers'?
Without profits, then new research will not be conducted by that company. I don't see anything more than that. Pricing is always what the market will bear, for any commodity.
> Ideally... I like to think that independent scientists would have the best ideas about where we should focus our attention on research / treatment. At the moment (as a researcher) one does need to alter what one wants to do (or recast it such that one gets funding for it). If there really was 'academic freedom' then... I think that would be a good thing, yeah. I know there are a number of scientists who work in university contexts for the explicit reason that yeah they don't make as much money as they could and yeah they don't have the nicest office or the flashiest equipment but if they develop the effective treatment for cancer or maleria they plan to write it up as a discovery for all to freely make use of rather than the discovery being patented by pharma (such that it won't be available to those who really need it).I think that's a pretty naive viewpoint. Prior to receiving funding from the public sector, there is always some sort of 'property rights' clause in the funding contract. Often, the institution holds those rights, or they share them with an agency of government. If funding is received from the private sector (most university research is funded by the private sector), then rights generally revert to the funder or are shared. Primary research outcomes, e.g. a new mechanism, end up being in the public domain. But any specific process arising from that new mechanism (or whatever) end up being patented. I can't think of an exception, outside of natural health products, which cannot be patented.
Even if that scenario of discovering a treatment for cancer or malaria would come to pass, there would still be huge expense in demonstrating all the things that would need to be demonstrated: safety, dose, efficacy. And then how would it be produced and distributed? Who would pay for that?
My observations of the real-life process run something like this.....Researcher makes fundamental discovery (public domain). Researcher develops specific approach to utilizing the discovery (patent). Researcher leaves academia and starts a company with private funding, or sells it to a private company outright, or the university does similar. Vast monies are spent in testing and development. There is a loss of some ideas at each step following the first, and maybe 1 in 1000 make it to the public.
> That doesn't surprise me. Its an pretty expensive medication, there is time remaining on patient, so seems like a good business decision to look into market expansion. (Don't get me wrong I'm sure there are a variety of different scientific hypotheses about the nature of hypertension and the nature of viagra that makes such a study scientifically interesting).
Actually, Viagra was first studied as a possible treatment for hypertension and angina. It didn't do so well at those things, but it was noted that penile erection was a ....errrr prominent side effect. The fact that the drug was being tested using standard methodology was what allowed this result to achieve notice.
Lar
poster:Larry Hoover
thread:896175
URL: http://www.dr-bob.org/babble/social/20090513/msgs/896425.html