All professional ethics are dependent on the epistemology of the profession. The possibility of following a code of ethics, whether the official one or an alternative code, is dependent on being able to obtain knowledge and understand the world. Professional knowledge has to be based on inferences from limited information. Statistics provides the optimal methods for making such inferences, and thus ethical professional conduct requires individual or collective understanding of some statistical thcory and practice. This is demonstrated using the medical (...) profession as an example. Statistical methods are essential in all areas of medicine, in routine medical practice, in screening for disease and in research. Evidence from medical journals, textbooks and research bodies shows that the statistical basis for ethical conduct is lacking. However, there are attempts to improve the situation. (shrink)

A patient's informed consent is required by the Nuremberg code, and its successors, before she can be entered into a clinical trial. However, concern has been expressed by both patients and professionals about the beneficial or detrimental effect on the patient of asking for her consent. We examine advantages and drawbacks of popular variations on consent, which might reduce the stress on patients at the point of illness. Both informed and uninformed responses to particular trials, and trials in general, are (...) discussed. The selection by doctors of patients, to whom entry to trials will be offered, is explored. (shrink)

This paper outlines the approaches of two apparently competing schools of statistics. The criticisms made by supporters of Bayesian statistics about conventional Frequentist statistics are explained, and the Bayesian claim that their method enables research into new treatments without the need for clinical trials is examined in detail. Several further important issues are considered, including: the use of historical controls and data routinely collected on patients; balance in randomised trials; the possibility of giving information to patients; patient choice and patient (...) autonomy; and how widely the results of clinical trials can be used. It is concluded that good statistical techniques in the design and analysis of medical studies are essential, but the statistical school used in developing such techniques is relatively unimportant. (shrink)