Power analysis in clinical trials

[deleted205891]

Advertisement - Members don't see this ad

Basically the main idea is that many drug trials are stopped when they shouldn't be just because of poor design, unable to detect a true difference

Do you think this topic is feasible for a semester-long independent research project? I am not sure, since there are many clinical trials, but this would be quite useful for pharmacy if it drew some good conclusions.

I trust that no one here will steal my idea since very few people on here dabble in hardcore statistics.

 

[Megas]

Poor design for power of a study can shut things down, but so can cost, especially for recombinant proteins during a trial--look at Xigris, dosing in the obese wasn't done during the clinical trials cause it's weight-based and would have just cost to much. Perhaps it would be easier to do this project if you limited it to non-recombinants or even non-bioligicals altogether (so no monoclonals either, maybe even exclude liposomal drugs) to simplify your one-semester project. That way you could focus on design of the study and not worry about cost-prohibitiveness of a study-in reality, this also can shut things don pretty fast.

 

[njac]

is it poor design or not enough participants to be adequately powered?

 

[ItsOverZyvox]

studies are meaningless now...all sponsored by pharma to show the results they want to show...with data sliced and diced and salamied.

No doctors ever read them... and effective marketing rules.

My job is to counter detail.

I am..the counter detail man.



Hmmmm... sounds like a silly super hero sitcom...

 

[deleted205891]

is it poor design or not enough participants to be adequately powered?

Well, sample size is part of the design.

Yes power is affected by sample size, both in the overall sample and in individual groups.

But I agree about cost-prohibitiveness. Since I'd be looking at real trials, maybe that wouldn't be that much of an issue. But I agree that to simplify the project I'd probably need to look at only one or two classes of drugs.

 

[Megas]

Well, sample size is part of the design.

Yes power is affected by sample size, both in the overall sample and in individual groups.

But I agree about cost-prohibitiveness. Since I'd be looking at real trials, maybe that wouldn't be that much of an issue. But I agree that to simplify the project I'd probably need to look at only one or two classes of drugs.

Clinical trials with biolgical agents are real trials.

 

[deleted205891]

Clinical trials with biolgical agents are real trials.

I know.

I am just saying that if they were done, they were probably not too expensive to be done. But who knows, I'm not too good with economics.

 

[Megas]

I know.

I am just saying that if they were done, they were probably not too expensive to be done. But who knows, I'm not too good with economics.

Yes I know that you know, that I knew.

 

[Farmercyst]

studies are meaningless now...all sponsored by pharma to show the results they want to show...with data sliced and diced and salamied.

No doctors ever read them... and effective marketing rules.

My job is to counter detail.

I am..the counter detail man.



Hmmmm... sounds like a silly super hero sitcom...

Starring Zpacksux and his high-powered lawn mower to shred Drug rep propaganda and save doctor's from themselves.

 

[RxWildcat]

eh, I'd leave that project to a biostatistician, but then again I hate stats so you may be asking the wrong guy.

 

[deleted205891]

eh, I'd leave that project to a biostatistician, but then again I hate stats so you may be asking the wrong guy.

It's supposed to be a project in biostats. It's to cap off my stat minor that I'm doing along with my PharmD. And so far I can't think of anything else. At first I wanted to do something about financial responsibility of insurance companies, but I have no idea how you could quantify that, even if one were able to obtain the data.

Whatever, if I can't figure out something good by myself before next semester starts, the prof. says she has something already in mind for me. It's just that I want to do something non-generic so it would look better for resume purposes and whatnot.

Edit: oh when I just typed the words non-generic, it made me think of bioequivalency between generics and nongenerics. But it has to be close for them to be allowed by the FDA...that won't work either so I revert to my previous idea.
If I somehow could find data for foreign production of similar drugs we use in the states (same mechanism and whatnot) and see if they work better, now that would be a project!