Doing RCTs of a treatment tells you if it works. Without random assignment (or some clever experimental design) tracking outcomes really doesn't tell you anything.