Considerations on the statistical design of the ATMI study

The ATMI study of prevention of mediastinitis after cardiac surgery uses th e total amount of therapeutic interventions measured by the TISS-28 score a s the primary endpoint. The study is designed as an adaptive group-sequenti al plan with three sequences, as experiences with the TISS-28 in cardiac in tensive care units are scarce, According to the Bauer-Kohne method the samp le sizes of the second and third sequences can be calculated in a data-driv en manner after the results of the previous interim analyses are available. Thus the first trial phase can be considered as an internal pilot study.