Diet and Child Behavior Problems: Fact or Fiction?

Eileen Cormier, PhD, RN; Jennifer Harrison Elder, PhD, RN, FAAN


Pediatr Nurs. 2007;33(2):138-143. 

In This Article

Common Dietary Interventions

Additive-free Diet

Originally introduced in 1975, Feingold hypothesized that food additives (in particular, artificial flavors and colors, and naturally occurring salicylates) were associated with learning disabilities and hyperactive behavior in some children. Feingold's claims gave rise to an avalanche of case reports and media attention, but controversy ensued in the scientific community about the validity of his findings, which were based on clinical observations rather than rigorous experimental evidence (Wender, 1986). Subsequent controlled studies either evaluated the behaviors of hyperactive children while on the Feingold diet, as compared to a placebo diet (dietary crossover designs), or investigated responses to specific food dye challenges (Lipton & Mayo, 1983).

In the early crossover studies of the Feingold diet, hyperactive children were randomly assigned to either the elimination diet or to a control diet, then crossed over to the other treatment condition. These studies generally used behavior ratings by parents and teachers and occasionally psychological tests of attention and impulsivity to measure outcomes, as opposed to physiological indicators. Several reviews (Mattes, 1983; Wender, 1986; Williams & Cram, 1978) and one meta-analysis (Kavale & Forness, 1983) have concluded that the Feingold diet is not an effective treatment for hyperactivity, highlighting a range of methodological challenges and flaws such as defining the study population, obtaining an adequate sample size, specifying a consistent diet, ensuring dietary compliance, developing an equivocal placebo control, using an adequate dose of artificial additive or coloring, and measuring appropriate outcomes in a standardized and precise manner. Even studies employing a double-blind, placebo-controlled design with a clearly defined study population, adequate sample size, and outcome measures demonstrated negative or at least ambiguous results (Connors, Goyette, Southwick, Lees, & Andrulonis, 1976; Harley, Ray, Tomasi, Eichman, Matthews, et al., 1978). Researchers who limited their investigations to evaluating child behavior responses to specific food dye challenges found only a small group of children who responded to some aspects of the Feingold diet (Silver, 1986).

The most recent studies of the effects of specific food additives and/or preservatives on child behavior have been conducted using double-blind, placebo-controlled, challenge crossover designs with children who have a history of atopy or parent-reported adverse reactions to food additives, but do not necessarily meet diagnostic criteria for ADHD. Selection bias continued to be a problem in these studies due to attrition, with possible overrepresentation by families interested in hyperactivity. Children did, however, demonstrate increased behavioral symptoms when challenged with artificial flavors, most notably tartrazine and preservatives such as calcium propionate (Bateman, Warner, Hutchison, Dean, Rowlandson, et al., 2004; Dengate & Ruben, 2002; Rowe & Rowe, 1994).

The recent meta-analysis conducted by Schab and Trinh (2004) also warrants mention. This analysis focused specifically on the effect of artificial food colors on hyperactivity rather than the Feingold Diet as a whole. Compared to previous meta-analysis by Kavale and Forness (1983), these authors employed hypotheses that were more explicit and rigorous, consisting of only double-blind placebo-controlled trials, including 6 trials conducted subsequent to the earlier analysis, and used statistical techniques that more richly exploit the advantages of crossover trials. The results of this meta-analysis support the hypothesis that artificial food colors can contribute to symptoms of childhood hyperactivity in some children.

Sugar Elimination Diet

The idea that foods containing sugar, mainly sucrose, might have an adverse effect on behavior was first hypothesized by Shannon (1922) and revisited by Randolph (1947) in his description of tension fatigue syndrome. Sucrose later appeared as a suspected offending substance in the 1970s as a result of coverage in the lay literature on the condition called functional reactive hypoglycemia (Deutsch, 1977). Using food diaries and observations of behavior, two cross-sectional studies found varying correlations between sugar intake and hyperactivity (Prinz, Roberts, & Hantman, 1980; Wolraich, 1996). These early studies were limited by the use of correlational designs in which it is impossible to determine causality or directionality. In short, it was as equally possible that the children's adverse behavior caused the increase in sucrose intake as it was that the increased sucrose intake caused the behavior. Other variables such as lack of parental discipline may have also been causal factors (Prinz & Riddle, 1986).

Wolraich, Wilson, and White (1995) conducted an extensive and thorough review of 16 double-blind, placebo-controlled studies evaluating the effects of sugar on child behavior. Participants included normal children, children identified by parents as behaving poorly after sugar ingestion, children with diagnosed hyperactivity or ADHD, and aggressive, delinquent children. Measures focused primarily on the behavior of children with ADHD and used behavior-rating scales completed by parents and teachers (along with neuropsychological measures) to assess vigilance, impulsivity, memory, and motor skills. In spite of considerable variation in subjects, challenge agents, and dependent measures, the results were remarkably consistent. Findings did not support the hypothesis that refined sugar affects hyperactivity, attention span, or cognitive performance of children, although the possibility of an effect on a subset of children could not be ruled out.

It is interesting to note that despite presentation of clinical evidence to the contrary, many participating parents remained convinced of an association between sugar and adverse behavior. White and Wolraich (1995) suggest that parental expectations may lead to mistaken interpretations about context-driven behavior variations (e.g., parties or holidays), associating them with sugar consumption. Hoover and Milich (1994) found that parents who believed their child was receiving a challenge dose of sugar, when it was actually artificial sweetener, rated the child's behavior as significantly worse and more demanding than parents who rightly expected their child to receive an artificial sweetener.

Food Allergies and Sensitivities

The idea that hyperactivity in children can result from sensitivity to specific provocative foods overlaps with existing conceptions of food allergies (Marshall, 1989). Several investigators have broadened Feingold's original hypothesis to restrict not only food additives and dyes, but also sugars, dairy products, wheat, corn, nuts, eggs, chocolate, and other foods that commonly cause allergic reactions in children (Boris & Mandel, 1994; Kaplan, McNicol, Conte, & Moghadam, 1989; Rapp, 1979). These studies have reported improvements in behavior symptoms associated with ADHD after 2-3 weeks on the experimental diet.

Other investigators have utilized more controlled research designs to assess the effects of the highly restrictive oligoantigenic diet (OAD), devoid of known food allergens (Carter et al., 1993; Egger, Carter, Graham, Gumley, & Soothill, 1985; Egger, Stolla, & McEwen, 1992; Schmidt, Mocks, Lay, Eisert, Fojkar, et al., 1997). OAD studies typically consist of three or four phases. In Phase 1, generally 4 weeks, subjects receive the OAD, with an alternative OAD available if they do not show improvement with the first version. Those subjects whose symptoms resolve by the last 2 weeks of Phase 1 move on to Phase 2, when excluded foods are gradually reintroduced. If symptoms do not recur, these foods are added to the subject's diet. Those who are eventually able to tolerate a satisfactory diet in terms of nutrition and preference, and who previously reacted adversely to a food for which both a test version and a placebo version are available, then enter Phase 3. This consists of the placebo-controlled, crossover food challenge trial, to assess whether symptoms are reproducible. The trial reintroduces one or more provoking foods or placebos, given for approximately 1 week, followed by a washout period. The studies cited above all reported improved behavior ratings both during the OAD period (Phase 1) as well as during the placebo-controlled food challenge (Phase 3).

As a whole, the OAD studies effectively demonstrate that food sensitivities or allergies can be involved in provoking behavior problems in certain children. They also suggest that children who do have identifiable sensitivities to certain foods might benefit from an elimination diet such as the OAD. However, evaluation of the relationship between food allergies and child behavior is complicated by the methodological problems in this line of research. These problems include (a) the questionable reliability and validity of various forms of allergy tests; (b) the small sample sizes, precluding the possibility of demonstrating significant results (only a small percentage of children enrolled in the OAD portion of the study eventually qualify for the food challenge trial phase); (c) the use of subjective rating scales from informants (usually parents) who may be biased; and (d) difficulties in determining appropriate challenge doses and timing of testing to accommodate the idiosyncrasies of different individuals' allergic reactions (Rojas & Chan, 2005; Marshall, 1989).

Fatty Acid Supplementation

A recent area of study involves essential fatty acids (EFA), in particular arachidonic acid (AA), eicosapentaenoic acid (EPA), and docosahexanoic acid (DHA). It is well established that these fatty acids are important in the structural and functional development and maintenance of neuronal membranes (Horrocks & Farooqui, 2004; Wainwright, 2002). Deficiencies in essential fatty acids have been implicated in the pathogenesis of a range of developmental and behavioral disorders, including ADHD (Burgess, Stevens, Zhang, & Peck, 2000; Richardson, 2004). Further, studies identifying lower plasma concentration levels of certain essential fatty acids among children with ADHD have led researchers to postulate that deficiencies are responsible for key features of ADHD (Colquhoun & Bunday, 1981: Stevens, Zentall, Abate, Kuczek, & Burgess, 1996; Yehudi, Rabinovitz, & Mostofsky, 1998).

To date, however, published trials of fatty acid supplementation in children with ADHD have failed to demonstrate improvements in symptoms of ADHD (Aman, Mitchell, & Turbott, 1987; Arnold, Kleylamp, & Votolato et al., 1989; Arnold, Kleykamp, Votolato, Gibson, & Horrocks, 1994; Richardson & Puri, 2002; Stevens et al., 2003; Voigt, Llorent, Jensen, Frayley, Berretta, et al., 2001). Again, a range of methodological problems may have limited the researchers' ability to identify true effects. Most of the studies had small sample sizes, did not confirm a diagnosis of ADHD, included subjects with coexisting psychiatric and developmental conditions, and addressed the issue of concurrent stimulant medication differently (Rojas & Chan, 2005). Refinements in study design are indicated before firm conclusions can be drawn about the efficacy of EFA supplementation with children who have developmental or behavioral disorders.

Gluten-free, Casein-free (GFCF) Diet

In recent years there has been mounting interest in the use of a GFCF diet for individuals with autism. Unfortunately, there is a paucity of empirical data to support claims that even include "miraculous cures." Cade and colleagues (1999), some of the first to study this diet in autistic populations, expanded initial laboratory work to applied settings and conducted a study of 270 individuals. One hundred and twenty of these participants were diagnosed with schizophrenia, and 149 met the DSM III criteria for a diagnosis of autism. All children with autism were treated with a GFCF diet, a synthesis of the Milk Free Kitchen by Kidder (1991) and the Gluten-Free Gourmet: Living Well Without Wheat by Hagman (1990). During the study, parents, physicians and some teachers independently assessed the children for the presence and severity of the diagnostic manifestations of autism using a 4-point Likert scale. These ratings were done initially and repeated after 1 month of treatment and then every 3 months for 1 year. Parent and physician reports were averaged, with variability of individual observer scores reported as less than 10%. Blood samples were examined to measure the absorption of peptides contained in wheat products (gluten) and dairy (casein) and the associated antibodies, immunoglobulin G (IgG) and transindolylacryloylglycine (IgA) for each of these food products. The study found that 87% of the children with autism had high titer IgG antibodies to gliadin, and 30% had high titer IgA antibodies to gluten or casein prior to initiation of the diet. Treatment with a GFCF diet was accompanied by reports of improvement in 81% of children within 3 months. A strength of this work was the combined use of physiological and behavioral measures. The behavioral results are limited, however, by the heavy reliance on reports from parents and teachers who knew that the children were on the GFCF diet.

In a comparative study, Arnold, Hyman, Mooney, and Kirby (2003) evaluated amino acid patterns of 26 children with autism on a regular diet, 10 on a gluten-casein free diet, and 26 with developmental delays who served as controls. The children with autism had higher deficiencies in essential amino acids compared to the control group. These findings suggest that children with autism are at high risk for amino acid deficiencies and may benefit from a structured diet. Clearly, this is an area that warrants further investigation. The authors note that a major limitation in the study was the small sample. An additional concern is the lack of strict dietary control for children on gluten-casein free diets, a commonly encountered problem in conducting dietary research in children.

Knivsberg, Reichelt, Hoien, and Nodland (2002) and others conducted a randomized single blind study with 20 subjects to assess the effect of a gluten-casein free diet on children with autistic syndrome and urinary peptide abnormalities. The children in the control and experimental groups were matched according to severity of autistic symptoms, age, and cognitive level. The experimental group showed more significant changes than did the control, and demonstrated improvement in autistic behavior, non-verbal cognitive level, and motor problems. Conversely, Elder and colleagues (2006) recently conducted a randomized clinical trial of the GFCF diet with 15 children diagnosed with autism and no statistically significant findings were noted on any of the objective measures, although several parents reported perceived improvement. Clearly, there is a need for replicating this work with larger samples and rigorous controlled clinical trials that evaluate both physiological and behavioral effects.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: