Rapid Responses to:

LETTERS: David L Sackett Turning a blind eye: Why we don't test for blindness at the end of our trials BMJ 2004; 328: 1136-a [Full text]

Rapid Responses: Submit a response to this article

Rapid Responses published:

A reason to be Bayesian?

Hans-Hermann Dubben   (2 September 2005)

A reason to be Bayesian? 2 September 2005
Hans-Hermann Dubben, Senior scientist Institut fuer Allgemeinmedizin - University Medical Center - 20246 Hamburg - Martinistrasse 52 Send response to journal: Re: A reason to be Bayesian?	I read with very great interest David Sackett’s letter. He concludes that a-priori-probabilities (or a-priori-hunches) of efficacy influence the test for successful blinding. I retain: apparently a-priori- probabilities of efficacy do exist. There are two items I want to point at. Item 1: A-priori-probabilities of efficacy are also closely related to a-posteriori-probabilities of efficacy. Thus, David Sackett provided an important example showing that Bayesian interpretation (rather than hypothesis testing) of a trial is not only necessary but also feasible. Item 2: A-priori-probabilities were not balanced in David Sackett’s example (and probably in many other trials as well). Is it than still ethical to conduct a randomised trial in which patients are allotted to treatment that is a-priori felt to be inferior? Reference: David L. Sackett: BMJ 2004;328:1136 (8 May), doi:10.1136/bmj.328.7448.1136-a Competing interests: None declared