You have a built in assumption in your statement that you haven't acknowledged: that the drug helps rather than harms. A new drug may help or it may harm.

If the drug turns out to harm people, a healthy person may be able to deal with it and recover. While a person that is barely hanging on to life may be pushed over the edge. There's no data to support a new drug raises the survival rate. It may in fact lower it from low to zero. During testing you generally want to kill as few people as possible.

Once more data has been collected on the drug, then the balance starts to shift toward giving it to at risk groups.