General modalities of analyses that have been used for ICD studies are revi
ewed. Published "typical" examples are briefly described. The historical co
hort method is exemplified with previously unpublished data from the Seattl
e Cardiac Arrest Survivor database. The AVID Study database is used to comp
are the results obtained from nonrandomized methodologies with randomized m
ethodologies. Particular issues related to the use of the ICD for example,
mode of death, inability to blind, selection practice, and treatment decisi
on times make this a natural pedagogic platform.