Does Medical Research Help Us Make Better Clinical Decisions?

Andrew J. Vickers, PhD


November 16, 2007

In This Article

What Do Clinical Trial Results Mean?

It might be a bit weird hearing this from a statistician, but when I read clinical trial results I often have no idea what they mean. Here is some recent news from oncology: "Elantra" improves survival in metastatic liver cancer (or is "Elantra" a type of car? I forget: the names all seem to blend together after a time). Here is the summary of the trial results:

"Elantra" treatment was associated with a median survival of 11.9 months versus 8.5 months for controls ( P = .002). The hazard ratio was 0.70. Two-year survival was 26% for "Elantra" versus 15% for standard care.

Fairly standard stuff...and almost totally useless in the clinic. Let's start with the hazard ratio. A hazard ratio of 0.70 means that at any randomly chosen time, patients taking the drug have about a 30% lower chance of dying the next day than do the controls. This is just about as difficult to explain to a patient as the P value (for example, try telling the patient, "The probability of a dataset at least this extreme under the null hypothesis is .002."). Moreover, the hazard ratio cannot possibly help a patient make a decision about whether or not to use "Elantra." To understand why, we need to think about a field of statistics known as "decision analysis."


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: