Please ILLUMINATE me on these questions

So, now that Pfizer has discontinued development of its star candidate, I have a couple of questions about this trial, mostly concerning the size of the trial.

Why 7500 patients per arm?

For example, this trial is sufficient to detect, with 95% confidence, an adverse event that has an occurrence rate of 1 in 2500 = 0.0004. Regulators want to see at least 1 in 500 from the whole development program for new families of drugs. Conservatism is fine and good, though this kind of data has been collected through post-marketing adverse event reporting and can be used, if needed, to discontinue marketing of a drug (e.g. Baycol).

On efficacy: this is sufficient to detect an effect size of 0.046 standard units with 90% power (the high end used in clinical trials). The standard deviation seems to be around 13 for males and 15 for females. So they’ve powered a trial to detect a difference of around 0.046*14*sqrt(2) = 0.7 g/dL of HDL cholesterol. Why do they need to detect a difference this small? Even without the mortality and cardiovascular adverse drug reactions, does the cost of taking this drug once a day merit a 0.7 g/dL increase in HDL cholesterol?

Now, Pfizer being the pharma giant that it is, I imagine it takes advantage of every discussion possible with the FDA to design its trials, and certainly the FDA had seen the design of this trial before it was executed. I’m very curious as to what was discussed when it came to sample size for this trial.

Technorati Tags: ,


3 Responses

  1. Arteries Unplugged II: Great Expectations…

    The last paragraph of an AP story about Pfizer’s withdrawal of torcetrapib awakened my inner statistician: According to Pfizer spokesman Paul Fitzhenry, 82 patients taking the combination of torcetrapib died, compared to 51 deaths in the arm of the s…

  2. You might find the 2004 NEJM article to be of utility in answering some of your questions. A second point concerns your statement on the minimal detectable difference for significance. Pfizer is not so stupid as to try and foist this paltry difference as being of clinical utility and even if they did, any physician buying in to such a story would and should be drawn and quartered.
    Althought here may be dupes in the medical community, the profession in general is well read and that would include Mark Twain’s take on the statistical crew, e.g., “Lies, damned lies and statistics”.

  3. And therein lies my confusion. Given that Pfizer is the largest pharma R&D company in the world, they have biostatisticians telling them exactly what they can detect with what power, and MDs telling them if anyone will care.

    I’ll have to check out the NEJM article in my spare time.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: