Authors' reply

BMJ 1996; 313 doi: https://doi.org/10.1136/bmj.313.7071.1550b (Published 14 December 1996) Cite this as: BMJ 1996;313:1550
  1. Stephen Sharp,
  2. Simon Thompson,
  3. Douglas Altman
  1. Research fellow in medical statistics London School of Hygiene and Tropical Medicine, London WC1E 7HT
  2. Professor of medical statistics and epidemiology Department of Medical Statistics and Evaluation, Royal Postgraduate Medical School, London W12 0NN
  3. Head Centre for Statistics in Medicine, Institute of Health Sciences, PO Box 777, Oxford OX3 7LF

    EDITOR,—We agree with Stephen Senn that analyses based on measurable patient characteristics rather than on underlying risk are preferable from both clinical and statistical viewpoints. Such analyses, however, require data on individual patients, which are not usually available to meta-analysts at present, while it may be possible …

    View Full Text

    Log in

    Log in through your institution


    * For online subscription