Replies to post #92412 on Biotech Values

[DewDiligence]

From ABT’s own PR on the ACCORD study:

http://finance.yahoo.com/news/Abbott-Statement-ACCORD-Lipid-prnews-851660241.html?x=0&.v=1

The average triglyceride level of a patient that starts fenofibrate [TriCor or TriLipix] in the United States is 302 mg/dL, according to medical claims data. The median triglyceride level in ACCORD Lipid was 162 mg/dL. The subgroup of patients in this study with triglycerides at or above 204 mg/dL reflects real-world fibrate usage.

…In a pre-specified subgroup of patients with both high triglycerides (at or above 204 mg/dL) and low HDL (at or below 34 mg/dL), there was a 31 percent lower rate of cardiovascular events in the fibrate-plus-statin arm compared to the statin-only arm (p = 0.057).

In other words, the 31% risk reduction in the “real world” fenofibrate-indicated subgroup missed the typical 0.05 definition of statsig by a whisker. (Whether a p-value of 0.05-epsilon would have been deemed statsig for this subset analysis is unclear from the information at hand but, ordinarily, success on a secondary endpoint or subgroup analysis is not given any standing statistically unless the primary endpoint has been met for the ITT group.)

[genisi]

Peter M. Nilsson, M.D., Ph.D. in the accompanying editorial on results from pre-specified group of patients with dyslipidemia:

In a prespecified subgroup analysis, there was a trend toward benefit of fenofibrate in patients with signs of dyslipidemia, which was defined as a triglyceride level of 204 mg per deciliter (2.30 mmol per liter) or more and an HDL cholesterol level of 34 mg per deciliter (0.88 mmol per liter) or less. This finding is of potential importance, since the ATP III guidelines define a high triglyceride level as 200 mg per deciliter (2.3 mmol per liter) or more and a low HDL cholesterol level as below 40 mg per deciliter (1.0 mmol per liter).2 Since many of the patients in this subgroup analysis did not meet these criteria, the role of fibrates for correcting dyslipidemia in high-risk patients with diabetes is still not settled.

http://content.nejm.org/cgi/content/full/NEJMe1002498?resourcetype=HWCIT