Royal Prince Alfred Hospital Royal Prince Alfred Hospital
Allergy Unit

Student research

Behavioral and cognitive effects
of food sensitivity in children:
towards a routine assessment protocol

Kaarina Sutinen
Master of Psychology, The University of Sydney
Supervisors: Stephanie Whitmont, Robert Loblay, Velencia Soutter, Anne Swain
January 1995

pdf Full Text - PDF (2,082 KB)


This study set out to assess whether measurable food effects on behaviour and cognition could be detected. A comprehensive review of research on food sensitivity provides tentative support for a link between certain food substances and hyperactive behaviour, impulsiveness and attention problems in children. However, there has been a dearth of experimentally rigorous research in this field based on the between-groups design. Most of the research has a number of defects, the most obvious being that there is a lack of appropriate control groups. In other words, most of the food challenge research has not used a control group of subjects who are in the normal range for symptoms of Attention-Deficit Hyperactivity Disorder (ADHD) to provide a benchmark for behavioural food reactors. Such research as exists did not use an identical dietary treatment for the control group. Hence, non-specific treatment effects could not be determined; and neither could the practice effects of repeated use of standard measures be controlled.

Other methodological weaknesses and technical problems in this research relate to subject selection, dietary and food challenge procedures and cognitive assessment. Some studies have "selected subjects to fit the test" on a post-hoc basis using the dependent measure, hence choosing the "best performers" and thereby confounding the dependent and independent variables. Furthermore, there have been widely different criteria by which problem behaviours are used to define subject samples, with some studies selecting subjects based on a broad spectrum of irritable, aggressive and oppositional behaviours. Other studies have not adequately eliminated foods that have been suggested to be common aggravators of behavioural reactions. Some of the earlier studies used diet replacement designs which make double-blind conditions difficult to achieve. A further major weakness has been that most studies have only assessed behavioural changes and not cognitive changes.

The present study employed a controlled experimental design using a double-blind cross-over (reversal) of active and placebo capsule challenges to compare the effects of food substances on two groups of matched food sensitive children. The experimental group of 16 behavioural reactors was selected on the basis that their clinical histories appeared to meet the DSM-III-R criteria for ADHD and that their behavioural ADHD symptoms increased with administration of known food substances. The control group consisted of 16 children with somatic symptoms of food sensitivity. Subjects were selected from an original pool of 363 children who attended the Royal Prince Alfred Hospital Allergy Consulting Service between 1984 and 1994. There was extensive matching of subjects for demographic and developmental characteristics in addition to food sensitivity variables.

In order to measure food effects, performance on behavioural ratings and cognitive tests following challenge with the active substances was compared to performance on the same measures following placebo challenge. Behavioural ratings consisted of the Conners' Parent Rating Scale and the experimenter's observations of ADHD behaviours during the two assessment visits. Three cognitive tests were used - the Continuous Performance Test, the Stroop Colour and Word Test and the Controlled Oral Word Association Test.

The results of this study provide support for significantly increased hyperactive, impulsive and inattentive behaviours following active food challenge conditions compared to placebo in the behavioural group but not in the control group. These significant food effect differences between groups were indicated by parents' ratings (the Conners' Hyperactivity Index score was significant at the  p=0.049 level) and by the experimenter's ratings of ADHD-type behaviours during assessment visits (p=0.024). The results also suggest there were clinically significant food effects for the behavioural group on the Hyperactivity Index. Their mean T score was nearly one standard deviation higher under the active food challenge condition (T score of 70.94) than under the placebo condition (T score of 79.37). There were no treatment order effects or any interaction effects which would have qualified these findings.

There appeared to be a dissociation between behavioural and cognitive effects of foods in that none of the cognitive tests detected significant food effect differences between groups. It remains to be investigated whether other tests or larger sample sizes are needed to detect cognitive changes. For the time being, however, the present results do not support the hypothesis that the specific kinds of cognitive functions which were investigated in this study are affected by foods.

Two subsidiary analyses were also conducted to make sure there were no consistent trends amongst more homogeneous groups. First, those eight behavioural children who were assessed by multiple raters to have experienced a positive reaction ("food reactors"), and their matched control subjects, were compared to those eight children who had less obvious reactions ("non reactors"), and their matches. The aim was to explore whether the non-significant cognitive effects of the main analysis would hold in this smaller but more homogeneous sample. The results were consistent with the main analysis, supporting the notion that, even in children who had severe behavioural reactions, no cognitive changes could be detected.

The second subsidiary analysis consisted of a comparison of 10 behavioural subjects who met clinical criteria for ADHD and 6 non-ADHD behavioural subjects. The aim was to assess whether these two groups showed similar food effects. The results were consistent overall with the main analysis and suggest that non-ADHD and ADHD subjects had similar adverse behavioural reactions to foods, in the absence of consistent cognitive changes. Further, the non-ADHD children's scores were elevated into the clinical range for the Hyperactivity Index during adverse food reactions (97th percentile) whereas they were in the normal range under placebo conditions (61st percentile).

Food and behaviour research went through an early phase of activity in the 1970's and is again becoming the focus of some interest. The present study is the first of its kind to demonstrate a behavioural food effect. It paves the way for further work in this area and, for the present, makes some recommendations for use of a rigorous "mixed" research design which involves use of appropriate control groups and includes assessment of each subject's food reactions as part of a carefully controlled elimination diet and double-blind placebo-controlled challenge protocol. It adds weight to the rationale of using diet as an adjunct to treatment in those children who have ADHD-type behavioural problems related to food sensitivity.