My argument is this: If you can't count on people to take a drug that will drastically reduce their likelihood of dying, how can you trust them to accurately determine the trade-off between efficacy and safety?
Most patients don't display the ability to figure out risks and trade-offs now (and relatively simple trade-offs at that). Do you think they'll be able to figure out what "45% increase in the risk of heart-attacks (41.2 - 48.8%, 95% confidence interval) in populations who have had a transient ischemic attack in the last 180 days, excluding those who have diagnosed atherosclerosis" means to them?