Intended for healthcare professionals

Letters

What is the prior probability of a proposed new treatment being superior to established treatments?

BMJ 1997; 314 doi: https://doi.org/10.1136/bmj.314.7073.74a (Published 04 January 1997) Cite this as: BMJ 1997;314:74
  1. Iain Chalmers, Directora
  1. a UK Cochrane Centre, NHS Research and Development Programme, Oxford OX2 7LG

    Editor–Terence Stephenson and David A Walker ask whether parents and children who are considering participating in a clinical trial of a new treatment for acute lymphoblastic leukaemia should be made aware that “for the past decade the new experimental treatments [studied in randomised trials] have given better outcomes.”1 In other words, hunches based on non-randomised evidence about which developments in the management of acute lymphoblastic leukaemia would turn out to be real advances have had an excellent track record recently. In the light of this, what should be one's prior belief about the likelihood that the next new treatment proposed will also represent an advance?

    This is an important general issue, but, as far as I am aware, it has only rarely been addressed systematically. How usual is the reported recent track record of new treatments for acute lymphoblastic leukaemia, and how likely is it that this successful run of new treatments is attributable to chance?

    My impression is that there is a professional and lay tendency to assume that most developments in health care are advances. It would have been helpful if Stephenson and Walker had presented (or referenced) the evidence from the leukaemia trials to which they refer because I have encountered only three other relevant analyses, all of which suggest that new treatments are as likely to be inferior as they are to be superior to existing alternatives (D Machin and M K B Parmar, and M Buyse and O Dalesio, colloquium on long term clinical trial strategies, Worcester College and Radcliffe Infirmary, Oxford, 15-17 December 1989).2 I would be grateful to anyone who could point me to other data, preferably derived from prospective cohort studies, that might help in an estimation of the prior probability of a proposed new treatment being superior to an established treatment.

    References

    1. 1.
    2. 2.
    View Abstract