Statistical designs in combination toxicology: a matter of choice.