The St. George's prospective treatment study of anorexia nervosa: A discussion of methodological problems