It's like saying that the theory of gravity is wrong because you droped a ball from a building and when you went to the street expecting to find the ball on the ground it wasn't there. Then everyone is like holy shit where is the ball? There is no evidence of it hitting the ground, say it was covered in paint to mark the exact point of impact. Too bad nobody saw the child who ran by and caught the ball in mid air before it hit the ground. The issue with trying to account for variables like this is simply that most of them are completely unrelated to the system that we are studying and infinite in their possibilities. There is a difference between a known parameter whose quantity can be random and multiple unknown parameters that are all random in every way that can possibly be conceived. This is what happens when large molecules adopt unexpected shapes, or unexpected enzymes or metabolites interact with a drug in the body due to diet, preexisting disease, or some other unknown factor that couldn't have possibly been predicted.
The use of chaos theory is common in electrochemistry as well as other areas of chemisry and science in general - Polynomial chaos in simulation and engineering applications F.
but it cannot be realistically applied to many problems in engineering, physics, chemistry, etc especially in situations where the products being made are used for manufacturing or in the human body. It really isn't possible to know what kinds of alterations in the parameters will be forced by the environment... It's like asking why we aren't using chaos theory to predict the exact time of death of every individual on the planet, or why we aren't using chaos theory to predict who will be elected for president 50 years from now. Chaos theory is not psychic, so there are limitations to what it can do for certain areas of research because the randomness is compounded and therefore not easily defined.
edit: so in short this is why we need experimentation in science so that we can a.) be sure of cause and effect and b.) be sure that the ideas we have and are testing are repeatable in all systems c.) and when there are exceptions we need to understand why (this is better done with experimentation because it has a way of revealing parameters that hundreds of years of theorizing and philosophizing would have never eluded to.