(6) Because valid dietary testing requires that a food substance be absorbed through the usual pathway at the usual rate, each food must be tested in the form in which it is normally consumed.

Based on these principles, Egger, Carter and Soothill developed a standard diagnostic protocol and used it for two trials. In the first they studied childhood migraine 45, in the second hyperkinetic syndrome/attention deficit disorder⁴⁶. All stages of the work were carried out while the children lived at home. The response of children to the diagnostic diet and reintroduction of foods were first determined by open experiment. The initial oligoantigenic diet was followed for four weeks. It consisted of one meat (chicken, lamb or turkey), one starch (potatoes or rice), one fruit (apples, pears or bananas), one vegetable from the brassica family, sunflower oil, a multivitamin, calcium and mineral water. The results were assessed by parents at home and by doctors during visits to the clinic. At the conclusion of the open phase of the trials, each child was considered to be food-intolerant if he remained symptom-free on the oligoantigenic diet and relapsed with addition of specific foods. To be included in the second phase, each child had to remain symptom-free by avoiding only those foods to which he was thought to be reactive. In phase two, the results of the open trial were tested in a double-blind, placebo-controlled cross-over experiment. One food to which the child had reacted in the open trial was consumed daily for a week, in a disguised form, indistinguishable from placebo, in quantities which the child would normally eat. The base in which the foods were hidden consisted of rice flour, carrot or banana, caramel, onion and salt or cane sugar and citric acid. Accuracy of blinding was assessed by the investigators. Only 5% of parents were able to correctly separate placebo and active challenge food by taste or smell. Most parents were unable to distinguish one substance from another and 12% guessed incorrectly.

The migraine study involved 40 boys and 48 girls, aged 3 to16, with headaches occurring at least once a week for six months to eleven years (mean 3.73 years), associated with two of the following symptoms: pallor, photophobia, dizziness, nausea, abdominal pain, visual disturbances or focal neurologic deficits. Classical migraine was the diagnosis in 39, common migraine in 49. During the open trial, 78 children (89%) became symptom-free and 4 children greatly improved. Relapse with refeeding of specific foods occurred in 90% of the responders. The interval between exposure and provocation varied from one hour to one week but averaged two to three days. The number of foods which provoked headache ranged from one to twenty-four. The frequency with which specific foods provoked headache is shown in Table 1. Forty children were selected for the double-blind placebo-controlled crossover study, the results of which are summarized in Table 2. This trial confirmed 65% of the food reactions identified in the open trial, using the strictest criteria available for clinical studies. Considering the total group of 88 children with severe and frequent migraine, 52% were shown to be intolerant of specific foods in this experiment. It is of note that when children were maintained on a dietary regime devoid of provoking foods, they were also resistant to other, non-specific triggers which had previously been thought to activate migraines, such as emotional distress, physical activity and temperature change.

When the trial of children with attention deficit disorder produced similarly dramatic results, the findings were challenged by Professor P. J. Graham in the Department of Psychiatry at Great Ormond Street and Dr. Eric Taylor, Head of Child Psychology at the Maudsley Hospital. A second study involving hyperkinetic children was instituted at Great Ormond Street, with the sceptical participation of Graham and Taylor, and the results of the first study were confirmed⁴⁷.

Table 2. Results of double-blind placebo-controlled
cross-over trial, 40 children, one food each
(Egger et al, Lancet 1983)