Lambrou, Dimitrios Nikolaos; (2001) Practical considerations in designing and analysing cross-over clinical trials. Doctoral thesis (Ph.D), UCL (University College London).

Text Practical_considerations_in_de.pdf Download (7MB)

Abstract

The problem of carry-over in cross-over trials has driven research activity for many decades. Depending on the assumptions made concerning carry-over effect in the 2×2 design, the CROS or the PAR estimator is selected for estimating treatment effect. The two stage procedure, selecting CROS with probability p and PAR with probability 1-p, achieves lower power and higher type I error-rate when compared to CROS. A corrected scheme, which achieves the nominal type I error-rate, proves inferior to alternative schemes regarding power and Mean Square Error Estimation rate. When baseline measurements are included in the analysis of the 2×2 design, a three-stage procedure emerges with similar properties to the two-stage one. The optimum plan for designing a cross-over study in families with more than two periods and/or sequences, depends on the assumptions made for the carry-over effects and the optimality criterion chosen. Best plans for two treatments, when model mis-specification occurs in both the systematic and/or random part of the model assumed to have generated the observed data, are derived. When three or more treatments are compared, optimum cyclic plans are chosen under a wide range of assumptions concerning carry-over activity.

Download activity - last month

Export as XMLCSVJSON

Download activity - last 12 months

Export as JSONCSVXML

Downloads by country - last 12 months

Export as XMLCSVJSON

Archive Staff Only

View Item