Biotech Values

[iwfal]

3) Poor response to Lipitor. Although you advance this as one source of bias, it's highly unlikely that patients would respond poorly to Lipitor and then respond to simva. It's even less likely that a physician would try simva in a poor Lipitor responder.

You've missed P3's point - if you respond with low efficacy to Atorvastatin then:

a) You are likely to respond with similarly poor efficacy to a drug in the same class. I.e. Simvastatin.

b) If you respond poorly to a second drug in the same class then you are likely to stop taking the drugs altogether.

I think the above scenario explains the known facts and pretty completely debunks the study - and, being the cynic that I have become, it wouldn't surprise me if Pfizer knew that they were taking advantage of a self-selection bias before they even started the study.