Re: [Fwd: CLIN PHAR STAT: Mixed Vs Fixed]

From: Stephen Senn <stephens@public-health.ucl.ac.uk> Date: Tue, 8 Aug 2000 11:59:19 GMT0 Subject: Re: [Fwd: CLIN PHAR STAT: Mixed Vs Fixed] Dear Nick, James and Mats Mats is right in saying > > Presumably, the severity of this problem decreases as the number of > > patients increases - in the context of meta-analysis, because here the patients are the repeated measurements on the trial effects. What we need to know is the correct variance from each trial and as the patients increase we get closer to estimating the true within-trial variance. I have a letter on this subject which will be appearing in Controlled Clinical Trials which shows that David Salsburg's recent contribution to that journal underestimates this problem and that he is wrong to recommend meta-analysis as an alternative to the linear model. I am very ignorant on PK/PD but the analogy here would seem to be not in the number of patients but in the number of measurements per patient. In this context, there may be a bias in variance estimation and associated inferential statistics (CI, P-values (ugh!) etc) for sparse sampling. However, it depends on the way you set the model up. If you impose a common residual variance for each patient then the problem largely disappears. (THis is analogous to what is done in a fixed effects linear model analysis of a multi-centre trial.) Of course, fully Bayesian methods put a prior on everything and so deal with this problem. (Or at least appear to deal with it.) Regards Stephen Professor Stephen Senn Department of Statistical Science & Department of Epidemiology and Public Health University College London Room 126, 1-19 Torrington Place LONDON WC1E 6BT Tel: +44 (0) 20 7679 1698 Fax: +44 (0) 20 7383 4703 Email: stephens@public-health.ucl.ac.uk webpage: http://www.ucl.ac.uk/~ucaksjs/