Clinicians need to know two things in order to make ethical decisions:
1) how likely is it that my patient will benefit or be harmed?
2) how big is the benefit or harm?
I analyze published data regarding the 3 kinds of treatments (preventive, curative and palliative) and 2 kinds of tests (screening and diagnostic) to illustrate that they rarely provide us with these two simple but vital pieces of information. I explain how researchers need to answer different questions than clinicians do and propose guidelines for the statistics that should be included in any publication of evidence intended for clinical use.