Dr David L. J. Freed, MB, MD, MIBiol
We all like to think that we know how to diagnose allergies, but do we really?
In truth, in my 30-plus years as an allergist, I've never learned the knack.
I'm only happy that I've got the diagnosis right when the patient recovers,
and even then I can't be sure. Perhaps he got better not because of my efforts but
despite them?
In my academic days I spent years developing a method of quantifying antibodies against foodstuffs, in the hope that it would be a useful diagnostic test. Janet Ditchfield, my PhD student, enrolled 24 migraine sufferers who were willing to try the diet indicated by the results. 12 went on the indicated diet. The other 12 went onto a "placebo diet" in which the foods to be avoided were randomly allocated. After that the two groups swapped over. The patients were not told which diet was based on the antibody test and which was placebo. In each phase, half of the patients improved, irrespective of whether they were on the "active" diet or the placebo diet. Since then, my requirement for any diagnostic test of food intolerance is that it should enable ill people to become well at least as often as a diet based on my own informed guesswork. This is called the 'utility value'. I confess I have not formally compared the utility values of Miller-type diagnosis and my own guesses, but I doubt if they would be much different. Maybe someone out there knows different?
It is a common misconception that you can validate a diagnostic test by double-blind trials. Wrong. Double-blind trials are for validating treatments. Double-blind challenges are for validating allergy diagnoses.
To validate any diagnostic test you have to compare its results with those of some other test, a test that you know generates the "true result". Skin-tests of the Miller type have certainly been compared with the results of double-blind challenges, and have been found to be woefully discordant. Of course, that does not prove that Miller-type skin-tests are no good for diagnosis, it only proves that they don't match with D/B challenge studies. Perhaps it is the latter that are at fault. Indeed, there are many theoretical objections to making the double-blind challenge our gold standard; it is probably far too insensitive and open to confounding factors(*).
So we end up not knowing. I can't prove that Miller-type testing does not give accurate diagnoses, and you can't prove that it does, because neither of us knows how to make a rock-solid diagnosis in the first place. We both agree (and so do conventional allergists) that some cases are easy - those are the ones who always react after exposure, quickly and convincingly with objective clinical signs. But we know also that some cases are much harder - patients for example who only react when certain combinations of foods are presented (see Casebook), or only if together with exertion or stress, or only premenstrually, or only after a lag of several days. What about those? It is precisely the claim of the old Clinical Ecologists to be able to make firm diagnoses in these cases that generated the heat, and caused our enemies to accuse us of manufacturing illusory illnesses. More seriously, the heat of the debate still blinds them to the real value of neutralisation as a method of desensitisation.
So what can we do, to find out if a certain diagnostic test is worth doing? We can work out the utility value - how many patients get better when they adopt the diet indicated by the relevant test, and how many get better when they go onto a standard few-foods or stone-age diet. If the "test' patients do better than the "standard diet" patients, we shall know that the test is worth doing, although we still won't know and will never know whether the test gives accurate results.
In summary; I do not deny that Miller-type testing might give accurate diagnostic information, but we have no way of knowing either way. Since that knowledge will probably never be obtainable, I work with what I've got, that is, a technique with which I can convert a satisfying number of ill people into well people. Why look for trouble with conventional colleagues when we don't need it?
Reference
*) Freed DLJ.
Laboratory diagnosis of food intolerance.
in Brostoff J, Challacombe S (eds)
'Food Allergy & Intolerance'
Eastbourne, Bailliere Tindall, 1987, 873-4.
