BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:The use of baseline covariates in cross-over studies - Mike Kenwar
 d\, London School of Hygiene and Tropical Medicine
DTSTART:20090519T133000Z
DTEND:20090519T143000Z
UID:TALK16130@talks.cam.ac.uk
CONTACT:Michael Sweeting
DESCRIPTION:In many settings\, cross-over trials that have within-period\n
 baseline measurements are analysed wrongly . A `conventional'\nanalysis of
  covariance in this setting uses each baseline as a\ncovariate for the fol
 lowing outcome variable in the same\nperiod\, but not for any other outcom
 e. If used with random\nsubject effects such an analysis leads to biased t
 reatment\ncomparisons\; this is an example of cross-level bias. Using a\np
 ostulated covariance structure that reflects the symmetry of\nthe cross-ov
 er setting\, such bias can be quantified and\, at the\nsame time\, potenti
 al gains and losses in efficiency through the\nuse of the baselines can be
  investigated. Alternative methods\nof analysis are described that avoid t
 he cross-level bias. The\ndevelopment is illustrated with two example tria
 ls\,\none balanced and orthogonal\, and one highly unbalanced and\nnon-ort
 hogonal. 
LOCATION:Large Seminar Room\, 1st Floor\, Institute of Public Health\, Uni
 versity Forvie Site\, Robinson Way\, Cambridge
END:VEVENT
END:VCALENDAR