Clark, I'm able to replicate your number for START, so I think our methods are very similar. Out of curiosity I cross checked this against Provenge's IMPACT trial which as you know also used OBF function. Trial info is given in the NEJM protocol document (section 1.12) as "Approximately 500 subjects will be enrolled (2:1 treatment vs. placebo) in order to obtain the approximately 304 deaths required for the final survival analysis. Overall survival: 88% power at a = 0.05, assuming a hazard ratio (HR) for death of 0.69"
What I get though is an alpha of .060. The difference is more than can be accounted for by rounding. As an additional check prior to the SPA amendment the protocol called for 360 events, 90% power, overall alpha .05, and HR 1.45 (1/.6897). When I run that through the formula I get alpha .041 (note Pocock boundary, interim at 50% info fraction).
I am puzzled at the large differences and as to what could be causing it. The only factor that isn't nailed down is the total number of events which use the qualifier "approximately", (e.g. IMPACT final was at 331 events) but the differences occur on both sides of .05 in the IMPACT example.
News 
Market Data 
Discover 
AI
