Well, it doesn't really solve the underlying pragmatic problems which have dogged 'allergen detection' efforts in foods all along.
MOST foodstuffs are not homogenous mixutres, and contamination
certainly isn't homogeneous.
Ergo, sampling is, INHERENTLY, flawed from the get-go.
All of the work that Dr. Taylor and colleagues do does not change that basic fact. I have
always objected to the notion that FARRP can 'clear' a food of contamination in the wake of a reaction to it.
"Well, not really," is my answer to that. They can
identify contamination if it is present in the sample evaluated at a level in excess of the limit of detection. Nothing more. That says NOTHING about contamination in a consumed sample, or in samples NOT tested.
To extend the conclusions of an analysis to samples not tested
assumes homogeneity among the population, or, at the very best, accounts for a lack of homogeneity via sampling methodology.
In other words, I'm not at all sure what the underlying PURPOSE of this research can possibly be for this particular group of researchers, because it certainly isn't what they seem to THINK it is.
Not only are they ignoring that most vulnerable 1% (or 0.1%, or 0.01% or whatever it is)... but they are also ignoring the physical REALITY of what food
is from a materials/chemical standpoint. It is NOT homogeneous (aside from a few exceptions such as oil or aqueous solutions without suspended particles), and even if it were, CONTAMINATION certainly isn't.
In short, this is NEVER going to absolve manufacturers of a responsibility to divulge processing to consumers with food allergies. Never. Because there is no realistic means of ever making sampling on a production line meaningful in the real world-- outliers
matter here, and it isn't "total" contamination spread over the population that is important-- it's concentration in EACH and EVERY individual mouthful and serving within the entire population coming off of that production line.
Come up with a way of evaluating THAT, and this might mean something. Well, to least sensitive 99.9% or whoever is above that cutoff point, anyway.