STATISTICAL ASPECTS OF THE DESIGN AND ANALYSIS OF STUDIES TO COMPARE HEMODIALYSIS MEMBRANES

Some aspects of the statistical design and analysis of short-term stud ies of haemodialysis membranes are reviewed. The potential benefits of crossover studies comparing more than two treatments are emphasized. Designs based on Latin squares are advocated and the importance of avo iding carryover effects by ensuring adequate time between successive t reatments is stressed. Studies of this type will usually record data s erially within each period, so appropriate graphical techniques and th e use of summary measures for the analysis of this type of data are co nsidered. An example of the analysis of a study, using MINITAB, is inc luded. Some issues of sample size estimation are discussed, and finall y some more advanced designs and methods of analysis are briefly consi dered.