New results on optimal prevention of risk averse agents