Concorde was a randomized trial to compare immediate and deferred zidovudin
e (AZT) treatment for persons with asymptomatic HIV infection. Alpha was a
randomized trial to compare the efficacy of two doses of didanosine (ddI),
in persons with symptomatic HN disease who were intolerant of zidovudine. T
he two trials overlapped in time, and a single Data and Safety Monitoring C
ommittee (DSMC) monitored accumulating data from both trials. This paper de
scribes the deliberations of the DSMC and its recommendations to the parent
steering committee during the course of each trial. In the discussion on C
oncorde, I draw attention to the problems of interpreting early survival da
ta and trends in surrogate markers. With Alpha, interest focusses on the co
nflicting considerations that arose in discussions about possible terminati
on at an interim stage. Control Clin Trials 1999;20:207-228 (C) Elsevier Sc
ience Inc. 1999.