INFORMATION TO USERS

This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer.

The quality of this reproduction is dependent upon the' quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction.

In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion.

Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand corner and continuing from left to right in equal sections with small overlaps. Each original is also photographed in one exposure and is included in reduced form at the back of the book.

Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6" x 9" black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order.

U-M-I University M:croftlms Ir~lerr~JI'Onal A Be!1 & Howe!' Ir,rorm3lior Corn..'ar'! 300 ~Jortr Zeeo Road ,l,nn Arbor MI48'06·1346 US;' 313 761-4700 800521.0600

Order Number 9300340

Behavioral effects of food: An exploratory study

Niino, James Shigeru, Ph.D.

University of Hawaii, 1992

V·M·I 300 N. Zeeb Rd. Ann Arbor, MI 48106

BEHAVIORAL EFFECTS OF FOOD:

AN EXPLORATORY STUDY

A DISSERTATION SUBMITTED TO THE GRADUATE DIVISION OF THE UNIVERSITY OF HAWAI'I IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF

DOCTOR OF PHILOSOPHY

IN

PSYCHOLOGY

AUGUST 1992

BY James S. Niino

Dissertation Committee: Richard A. Dubanoski, Chairperson Daniel D. Blaine Thomas J. Ciborowski Walter Nunokawa Donald B. Char, M.D. @ Copyright 1992

by James shigeru Niino

iii ABSTRACT

A careful review of the vast literature on the effects of food and diet revealed some evidence of advarse reactions to food and food ingredients. There is, too, an increasing body of literature indicating that stress may be an important, if not necessary, precipitating factor in the expression of symptoms in various illnesses.

A double-blind, time series procedure utilizing two teenage male wards at a juvenile detention facility was designed to test for behavioral effects and assess the influence of stress on reactions. Two suspected offending foods were identified for each subject. The results were mixed. The data obtained included changes in levels of fidgeting in response to one food for each subject. The value of this study is that it tested a procedure for investigating behavioral reactions to food. Refinements in the design to improve its usefulness are discussed.

iv TABLE OF CONTENTS Abstract ••• •••.•••• . .. iv List of Tables •• ••• •• viii List of Figures •••••• ix

Chapter 1: Introduction 1

Feingold K-P Diet and Hyperactivity • 14

Clinical Ecology. • . •••. 61

Food Allergy. •••.••• ..• 73

Sugar Studies ••• •. •••••• 79

Diet and Antisocial Behavior. . •••. 92

In vitro and Animal Studies ••..•• 103

Stress and Illness ••••••• 107

Hypotheses. •..•••.•.•..• 131

Chapter 2: Method. •. . •.•••. 133

Design ...... 133

SUbjects .••••••••• 135

Apparatus ...... 137

Reaction Time. .••. 137 Blood Pressure/Pulse ••••. 138 Time Estimation. • •••. 139

Setting •••••• 139

Foods and Placebo • 140

Procedure 142 Dependent Measures ••••••. 142 SUbject Expectations and Guesses 145 Stress ••••• .... 145 v Practice Sessions...... 147

Baseline/Stress Verification . 148

Elimination Diet .....•• 149

Verification of Intolerance. . 151

Double-Blind Challenges. . 152

Chapter 3: Results. 154

Autocorrelation •. 154

Error Rates • 156

Measures. •• 156 Hypothesis 1 · 157 Hypothesis 2 · 164 Verification of Stress. . 176 Hypothesis 3 · 179 Hypothesis 4 180 Hypothesis 5 · 184 Hypothesis 6 · . . 196 Blood Pressure and Pulse Rate • 206

Subjects' Expectations and Guesses. 208

Chapter 4: Discussion. 211

Review•• 211

Design. 221

Baseline 221 Elimination Diet . 222 Foods. 222 stress • 224 Measures • 225 vi Future Research · · ·· · · · ·· · . . · · ·· 227 Appendix A: Medical History Questionnaire - Part I- · · ·· · · · ··· .. ···· 230 Appendix B: Medical History Questionnaire - Part II · · ·· · · · ··· · · · · 231 References ..... · 232

vii LIST OF TABLES Table

1. Definitions of Food allergy and Related Terminology ••••• 3

2. Offending Foods 63

3. symptoms Reproduced in Patients by Testing for Allergies •••••••• 64

4. Typical Adverse Reactions to Foods. 65

5. Mean Time Spent Fidgeting Under Stress and Nonstress Conditions ••••••• ••• 177

6. Pulse Rates After Stress and Nonstress Conditions . · · · · · · · · · · · · 178 7. Lowest Mean Reaction Time in each Condition for Subject 1 · · · . . . . ···· · · ··· 181 8. Lowest Mean Reaction Time in each Condition for SUbject 2 · · · . . . . ······ · · · 182

viii LIST OF FIGURES Figure

1. Reaction Time During ELIM I and ELIM II for SUbject 1 • • • • . • • 158

2. Fidgeting During ELIM I and ELIM II for SUbject 1 • • • • . • • • • 160

3. Reaction Time During ELIM I and ELIM II for SUbject 2 • • • • . • • 161

4. Fidgeting During ELIM I and ELIM II for SUbject 2 • • • • . • • . 162

5. Daily Means of the Time Estimation Trials for SUbject 1 · · · · · · · ··· ·· · 165 6. Daily Means of the Time Estimation Trials for SUbject 2 · · · · · ·· · ·· · · · 166 7. Fidgeting During Challenge Phases for SUbject 1 • • • • . • 167

8. Reaction Time During Challenge Phases for SUbject 1 ·· · ··· · ··· · 169 9. Reaction Time During Nonchallenge Phases for SUbject 1 ··· · · · · · ·· · · · 170 10. Fidgeting During Nonchallenge Phases for SUbject 1••••.••••• 171

11. Reaction Time During Challenge Phases for SUbj ect 2 • • • • . • . • 172

12. Fidgeting During Challenge Phases for sUbject 2 • • • • . • • • 173

13. Reaction Time During Nonchallenge Phases for SUbject 2 • • • • . • • •. 174

14. Fidgeting During Nonchallenge Phases for.Subject 2 ••••.••••• 175

15. Comparisons to Reaction Time Under the Stress Condition During Baseline for SUbject 1••...••••.• 186

ix 16. Comparisons to Reaction Time Under the Nonstress Condition During Baseline for Subject 1 • • . • . . • • . 187

17. comparisons to Fidgeting Under the stress Condition During Baseline for SUbject 1••.•••••.• 189

18. Comparisons to Fidgeting Under the Nonstress Condition During Baseline for SUbject 1••.••.••... 190

19. comparisons to Reaction Time Under the stress Condition During Baseline for SUbject 2•.••••••.•• 191

20. comparisons to Reaction Time Under the Nonstress Condition During Baseline for SUbject 2 . • . • • • • • • 192

21. Comparisons to Fidgeting Under the stress Condition During Baseline for SUbject 2•••••.•••• 194

22. Comparisons to Fidgeting Under the Nonstress Condition During Baseline for SUbject 2.•••••••••• 195

23. comparisons to Reaction Time Under the stress Condition During Baseline for SUbject 1.•..••••••• 197

24. Comparisons to Reaction Tima Under the Nonstress Condition During Baseline for SUbject 1 • . • • • • • • • 198

25. Comparisons to Fidgeting Under the stress Condition During Baseline for SUbject 1•••...•••• 199

26. Comparisons to Fidgeting Under the Nonstress Condition During Baseline for SUbject 1...•••••••• 200

27. Comparisons to Reaction Time Under the Stress Condition During Baseline for SUbject 2•.•.•.••.•• 202

28. comparisons to Reaction Time Under the Nonstress Condition During Baseline for Subject 2 . .• .•• ... 203 x 29. comparisons to Fidgeting Under the stress Condition During Baseline for SUbject 2 • • • • • • • • . • 204 30. Comparisons to Fidgeting Under the Nonstress Condition During Baseline for SUbject 2 • • . . • • • . 205

3l. Daily Differences in Pulse Rates for Subject 1 ...... 207 32. Daily Differences in Pulse Rates for Subject 2 ...... 209

xi 1 CHAPTER 1

INTRODUCTION

The realization that food can adversely affect susceptible individuals is by no means new. For example, the Greeks recognized idiosyncratic reactions to shellfish, shrimp, rhubarb, buckwheat, and cashews (Randolph & Moss,

1980). Hippocrates stressed how important it is for physicians to understand the totality of effects of food and drink and recorded observations of unpleasant reactions

(e.g., fatigue, mental dullness) to different foods

(Dickey, 1976b). Lucretius, a Roman poet, also wrote in about 60 B.C., "One man's meat is another man's poison"

(Truswell, 1985).

One of the problems that becomes immediately evident when reviewing the literature in this area is that there is no clear, common language. This is probably due to the complex nature of adverse reactions to food. Many have recommended that the various terms that are often used interchangeably, be precisely defined and used in a consistent manner in order to facilitate meaningful dialogue (Bock, 1980, 1984; Businco, Benincori, & Cantani,

1984, 1986; Coombs, 1984; Hanson, 1984; May, 1975, 1982,

1984, 1985; Metcalfe, 1984; Panush & Webster, 1985; Pearson & McKee, 1985; Pratt, 1958; Taylor, 1986; Truswell, 1985).

The definitions proposed by the American Academy of Allergy and Immunology Committee on Adverse Reactions to Foods as 2 presented in Anderson (1986) appears in Table 1. This terminology was incorporated in this paper. Most of the authors above agree with these definitions. Some, however, prefer the term food sensitivity rather than hypersensitivity when referring to the immunologically-mediated reactions (e.g., Bock, 1980; May, 1984; Panush & Webster, 1985; Taylor, 1986). Coombs and Gell (1963) have also delineated sUb-types of immunologic reactions (i.e., Types I to IV) according to whether the antibodies or antigens are cell-fixed or in solution and the immediacy of the reactions. Additional categories of reactions have also been described. These include food aversion or psychological intolerance referring to adverse effects of psychological origin which are not reproducible under blind conditions (Cant, 1985; Lessof, 1984; May, 1984; Pearson & McKee, 1985; Truswell, 1985). Lessof (1983) proposed that foods may have an irritant effect on already diseased tissue as is the case when one with an intestinal disorder eats a highly spiced meal and aggravates the vulnerable gut. Several have also described exercise-induced reactions (e.g., Maulitz, Pratt, & Shockett, 1979; McNeil & Strauss, 1988; Nevey, Fairshter, Salness, Simon, & Curd, 1983). In these cases, it appears that attacks only occur when vigorous exercise follows within a few hours of the food ingestion. This type of reaction is not well understood as 3

Table 1. Definitions of Food Allergy and Related Terminology Adverse reaction (sensitivity to a food) A general term that can be applied to a clinically abnormal response attributed to an exposure to a food or food additive. Food allergy (hypersensitivity) An immunologic reaction resulting from the ingestion of a food or food additive; this reaction occurs only in some patients, may occur after only a small amount of the substance is ingested, and is unrelated to any physiologic effect of the food or food additive; to many, the terms "food allergy and hypersensitivity" are syncnymous with reactions that involve an IgE-immunologic mechanism, of which anaphylaxis is the classic example; to others the term may include any food reaction known to involve an immune mechanism; these are overused terms that have been incorrectly applied to any and all adverse reactions to a food or food additive. Food anaphylaxis A classic allergic (hypersensitivity) reaction to food or food additives in which the immunologic activity if IgE homocytotropic antibody and release of chemical mediators are involved. Food intolerance Ageneral term describing an abnormal physiologic response to an ingested food or food additive; this reaction is not proved to be immunologic and can include idiosyncratic, metabolic, pharmacologic, or toxic responses to food or food additives; the term is often overused and, like the term "food allergy" (hypersensitivity), has been applied incorrectly to any or all adverse reactions to foods. Food toxicity (poisoning) A term used to imply an adverse effect caused by the direct action of a food or food additive on the host recipient without the involvement of immune mechanisms: this type of reaction may involve nonimmune release of chemical mediators; toxins may be contained within food or released by microorganisms or parasites contaminating products; on some occasions, the term may be synonymous with idiosyncratic adverse reaction; when the reaction is anaphylaxis-like, it may be called "anaphylactoid." Food idiosyncrasy A quantitatively abnormal response to a food substance or additive: this reaction differs from its physiologic or phar~acologic effect and resembles allergy (hypersensitivity) but does not involve immune mechanisms; food idiosyncratic reactions include those that occur in specific groups of individuals who may. be genetically predisposed; when the reaction is anaphylaxis-like, it may be called "anaphylactoid." Anaphylactoid reaction to a food An anaphylaxis-like reaction to food or food additive presumed to result from a nonimmune release of chemical mediators; this reaction mimics the symptoms of food allergy (hypersensitivity). Pharmacologic food reaction An adverse reaction to a food or food additive as a result of a naturally derived or added chemical that produces a druglike or pharmacologic effect in the host. Metabolic food reaction An adverse reaction to a food or food additive as the result of the effect of the substance on the metabolism of the host reci"ient. 4 it is a rare condition and there are very few opportunities to study it. Many agree that between 15% and 20% of the general population suffer from respiratory or cutaneous allergies (e.g., Barkin & McGovern, 1966; Butkus & Mahan, 1986; Gerrard, Ko, Vickers, & Gerrard, 1976; Mayron, 1979; Rapp, 1978). There is much less agreement in regard to the incidence of adverse reactions to foods. The estimates vary because of sampling methods and instruments (Lessof, 1988), diversity of symptoms, and, of course, the varying definitions. Estimates of incidence of food allergy range from .3% to 38% in the pediatric population (Bock & Martin, 1983; Collins-Williams, 1956; Freier & Kletter, 1970; Heiner, Wilson, & Lahey, 1964; Metcalfe, 1984; Moore, 1954). Estimates for the adult population range from less than 2% to 33% (e.g., Bender & Matthews, 1981; Burn & Merret, 1983; Gallagher, Riehm, .Valanis, & Bernstein, 1983; Yen, 1987). Breneman (1985) estimated that in the U.S., 30 million individuals suffer from this type of allergy. The American College of Allergists estimated that 25% of Americans (56 million) are affected by "true allergies" and that half (28 million) are food-related (Fink, 1984). The Department of Agriculture in the Federal Register (1983) reported that 15% of the population (34 million) may be sensitive to some food ingredients. 5 A prominent clinical ecologist, Theron G. Randolph, feels that food allergies or addictions and intolerances are one of the greatest health problems in the United states (Randolph & Moss, 1980). Another well-known ecologist, Marshall Mandell, estimated that between 50% and

80% of the patients seen by many physicians are suffering from reactions to foods and chemicals that they are sensitive to (Mandell & Scanlon, 1979) and that 75% of the general population are troubled by food allergies and/or intolerances (Mandell, 1981). Hosen (1988) reported that of 1,000 patients that he admitted to the hospital, 50% had combined inhalant and food allergies and 25% had the food allergy alone. Coca (1953) estimated that as many as 90% of all Americans are sensitive to one or more foods.

Although there is no consensus regarding the incidence of food allergy and intolerance, many would agree that reactions are more prevalent in infants and children than adults (Bahna, 1987; Bock, 1987; Breneman, 1985; Buckley & Metcalfe, 1982; Butkus & Mahan, 1986; Faelton, 1983; Fries, 1959; Minford, MacDonald, & Littlewood, 1982; Moneret­ vautrin, 1986; Panush & Webster, 1985; Parker, Sussman, & Krondl, 1988; Schreiber & Walker, 1989).

Although it can be argued that food allergy and intolerance are underdiagnosed (Minford, MacDonald, &

Littlewood, 1982), even the more conservative estimates indicate that adverse reactions to food constitute a 6 substantial problem. It is apparent, despite this range of estimates, that a "significant number of Americans may be at risk" (Sloan & Powers, 1986). It appears that more than just an unfortunate few react to things they eat. From the turn of the century, reports of physical and behavioral reactions to food have appeared in the medical literature from specialties such as allergy and pediatrics. For example, Schofield (1908) reported that one of his patients, a 13-year-old boy, reacted with urticarial rash, pUffy eyelids, itching, swelling, and asthma-like attacks whenever he ate eggs in any form. He supposedly even reacted after eating bacon that had been fried with eggs. Schloss (1911) wrote of an eight-year-old boy's reactions to eggs, almonds, and oats. Again, the symptoms were rash and facial swelling. Campbell (1927) published a follow-up of 77 allergic children and observations on 70 additional sUbjects. Among the cases reported were those of a 15­ month-old who reacted with eczema to lamb and the whites of eggs, and a 10-month-old who suffered an acute asthma attack after consuming an egg. Finally, Hopkins, Waters, and Kester (1931) reported that in their sample of over 100 children sUffering from eczema, food was the sole cause or a contributing factor in about a half of the cases with milk, egg, and wheat being the most common offenders. Many other cases are reviewed in Rowe's (1931, 1937, 1972) very thorough texts on allergy. Anderson & Sogn (1984), 7 Businco, Benincori, and cantani (1986), Cant (1985), Lessof (1984), and Pearson and McKee (1985) provide current reviews of the classic (i.e., gastrointestinal, respiratory, and cutaneous) and other somatic symptoms that have been linked to foods. The animal literature also contains some interesting case reports. For example, an article appeared in the Journal of the American Veterinary Medicine Association of a siamese cat that suffered from food hypersensitivity (Medleau, Latimer, & Duncan, 1986). This feline apparently experienced an intense form of pruritis (itching) as was evidenced by self-inflicted erythema (inflammation and redness of the skin) and excoriations. A long course of treatment with the usual medication (e.g., prednisolone, an analogue of cortisol) proved ineffective. A hypoallergenic diet consisting of canned lamb and rice was then tried and in 2 days she was no longer pruritic. Four days later, after all erosions and excoriations had healed, her regular brand of commercial canned cat food was reintroduced. In 4 hours she became again "intensely pruritic". She was returned to the low allergenicity diet and a follow-up 11 months later revealed no further bouts of pruritis. Reports of behavioral effects are less common. Two possible reasons are: (l) behavior is not always as definitive and quantifiable as physical reactions, and (2) most physicians were (and still are) concerned primarily 8 with the classic symptoms. A statement by Feingold (1975a) before he became interested in behavioral symptoms reflects this orientation. He said that he was an "allergist not a behaviorist" and that although alleviation of behavioral symptoms came to his attention, he did not emphasize them or investigate further, for it was not within the scope of the Allergy Department. Of the published behavioral accounts, the following are the most frequently cited. Hoobler (1916) in an attempt at collecting and categorizing what he called the manifestations of anaphylaxis listed irritability, restlessness, and fretfulness under symptoms relating to the nervous system. Shannon (1922) presented case studies of 8 patients who showed neuropathic diathesis with symptoms such as extreme nervousness, irritability, and unruliness. He claimed that "all showed definite relief from the nervous symptoms on the specific therapy directed at the proteins to which they were sensitive. In all but 1 of the 8 cases, the protein concerned were those foods contained in the dietaries of the patients" (p. 94). Rowe (1930), in a presentation to the Association for the study of Allergy, reviewed his statistical study of a hundred cases of gastrointestinal food allergy and reported that 10% of the cases experienced irritability, 12% nervousness, and 10% mental dullness and confusion. Alvarez and Hinshaw (1935) similarly found that "symptoms such as irritability, 9 sleepiness, dopiness, dizziness, numbness, queer feelings, cold sweats, feverishness, and perhaps some mental abberation" can result for many people after eating certain foods (p. 2056). They declared that this should be

"forcibly brought to the attention of the medical profession" (p. 2056).

Alvarez (1946) reported the case of a 41-year-old woman who experienced, in addition to many somatic symptoms, great anxiety and fatigue for 15 years or more.

These symptoms abated when she managed to completely avoid all milk products. Accidentally eating a dessert containing milk resulted in a recurrence of the symptoms.

Davison (1949) presented a summary of 87 cases of what he called cerebral allergy. Symptoms displayed by these patients included "definite mental and emotional disturbances, personality changes, mechanical and neurologic disturbances, disturbances about the eyes, and those of sight, and speech changes" (p. 713). He claimed that 50 of the 87 sUbjects had clinically proven food sensitivity. The ten most common offenders were milk, chocolate, onions, cabbage, pork, eggs, fish and shellfish, tomatoes, nuts, and apples. Many other case studies in which some form of undesirable behavior is linked to common foods can be found in Clarke (1950), Clein (1954), and

Speer (1954a, b). Reports of adverse behavioral reactions to foods continued to appear quite regularly in the medical 10 literature (Dees, 1959; Crook, Harrison, Crawford, & Emerson, 1961; Speer, 1963; RUbin, Shapiro, Muehlbauer, & Grolnick, 1965; Kwok, 1968; Wilken-Jensen & Melchior, 1970; Campbell, 1973; Speer, 1975; Fand & Hubbard, 1976; Singh & Kay, 1976; Finn & Cohen, 1978; Lessof, Wraiths, Merrett, Merrett, & Buisseret, 1980; Denman, 1980; Minford, MacDonald, Littlewood, 1982; Egger, Carter, Wilson, Turner, & Soothill, 1983; Businco, Benincori, & Cantani, 1984; Breneman, 1985; Settipane, 1986; Young, Patel, Stoneham, Rona & Wilkinson, 1987). Although these observations of physical and behavioral reaction to foods have been reported for years, as pointed out by many (e.g., Finn & Cohen, 1978), historically the concepts of food allergy and intolerance were, for the most part, not incorporated into the mainstream of medicine. Dickey (1976b) wrote: "To many physicians, food allergy is well within the realm of fantasy, except for the occasional patient with a fixed food reaction to an infrequently consumed food • The subject is not taught in medical schools ••• The physician who stresses nutrition and the food factors in disease is often looked upon as a food faddist" (p. 26). Some of the resistance and neglect in looking seriously at foods as etiology in physical and behavioral disorders may have been due to the following: 11 1. The focus of the field of orthodox allergy and medicine in general was (and still is) "best referred to as endogeny, consists principally of anthropocentrically focused immunologic and other analytic approaches. Clinical manifestations are usually treated nonspecifically and largely by means of drugs" (Randolph, 1976a, p. 11). To the traditional allergist, if there was no readily demonstrable immunologic etiology (e.g., evidence of reagins) for the symptoms, it was not an allergy. Many reactions to food and chemicals have no demonstrable immunologic basis (Coca, 1942; Knicker, 1987), so these were considered out-of-bounds or beyond the scope of study and largely ignored. Butkus and Mahan (1986), Check (1983), Crook (1975), Dickey (1976b), Lancet (1979), May (1975, 1980, 1984), Rossiter (1985), and Taylor (1986) all presented excellent discussions of this and related issues. Most of the medical profession accepted this narrow definition and conducted their practices and research within these boundaries. Attention was, therefore, focused primarily on molds, pollens, danders, dust, and a few foods (Faelten, 1983). Randolph (1976a) wrote also that endogeny is the dominant approach because it's more applicable enmasse and more economical short-term. Exogeny, on the other hand, is a challenge to teach and difficult and time consuming to apply. The endogenous approach is perpetuated, for it is the approach that is most often 12 taught in medical schools. others have noted how food allergy and intolerance have received only brief, if any, coverage in standard medical textbooks (Finn & Cohen, 1978; Crook et al., 1961). 2. "New truths are resisted because they disturb the status quo" (Mandell, 1976, p. 614). "New concepts can necessitate the altering of well-established routine procedures, and it is therefore not surprising that it disturbs one's thinking. It is well known that change is countered by a definite element of inertia. Thus, the fear of what is new, neophobia, is a real factor which has to be considered" (Dickey, 1976, p. 33). 3. For obvious financial reasons, there has been opposition to the idea of adverse reactions to food by segments of the food industry (Feingold, 1977a; Randolph & Moss, 1980). It has been claimed that attempts have been made by the industry to downplay or discredit reports of adverse reactions. 4. It was (and still is) commonly believed that, generally, all foods are safe for all. There was (and still is) disbelief that adverse reactions can be caused by something as necessary, "natural" and frequently consumed as foods. 5. Reports of reactions to food were (and still are) scattered and, as discussed earlier, the terminology (e.g., 13 food sensitivity, allergy, intolerance, anaphylaxis, etc.) is ambiguous and varied (Crook et al., 1961). 6. Finally, as noted by May (1975, 1980, 1984) and Check (1983), one of the most important reasons is that there was little empirical support for claims in this area because of the lack of adequately designed studies. The literature consisted primarily of clinical observations and uncontrolled case studies. A likely reason for this is the significant problems inherent in conducting studies on food. For example, double-blind conditions are difficult to arrange and maintain and preventing cheating on prescribed diets are oftentimes impossible without extraordinary measures. Although there have been some better designed studies, these were not free of legitimate criticism and, moreover, the results obtained were mixed. Currently this is still largely true in regard to behavioral effects. Any conclusions at this time would, therefore, be premature. The purpose of this study was to investigate the re~ationship between food and behavior with a carefully designed procedure. The studies mentioned above as well as those that will be discussed at length next provide some evidence that foods do impact behavior. These have important implications in regard to our understanding of behavior and research in this area needs to be pursued more vigorously and empirically. 14 Feingold K-P Diet and Hyperactivity During the last decade, there has been a significantincrease in attention paid to the possibility of relationships between food and behavior in both the lay and professional sectors. Much of this interest was stimulated by the introduction and tremendous popularity of the Feingold K-P diet for children with hyperkinesis-learning disability (H-LD). In managing aspirin-sensitive patients during the 1960's, Feingold discovered that favorable responses could be obtained in many more cases when all foods containing natural or synthetic salicylate radicals were also eliminated from their diets. At this point, then, his diet eliminated aspirin, some fruits and vegetables (natural salicylate radicals) and seven synthetic flavor chemicals (synthetic salicylate radicals). Despite many successes with this diet, a number of aspirin-sensitive patients still failed to improve. Medical reports then began appearing indicating that some individuals who experience adverse reactions to salicylate compounds also react to tartrazine (FD&C - "Yellow 5"), a chemical structurally unrelated to acetylsalicylate acid (e.g., Lockey, 1959, 1975, 1977). As reported in Feingold (1975a), via his own experience he discovered that the reverse could also be true. Observations were also published by Vane (1971) and Ferreira (1971) indicating that another drug, indomethacin, 15 which is also structurally unrelated to aspirin can produce reactions in aspirin-sensitive individuals. Because of these observed cross-reactions between structurally unrelated chemicals, Feingold (1975b) hypothesized that "among the several thousand colors and synthetic flavors in the food and drug supply, there might be other chemicals which, although not related structurally, might induce identical clinical responses" (p. 802). He then revised his diet to also eliminate all foods contain:.ng synthetic colors and flavors (Feingold, 1973a, bi 1975a, b; 1976; 1977a, b). He reported that with this diet, "although responses were not 100 percent on each occasion • • • we were overwhelmingly successful in patient management" (1975a, p , 9). Up until the late sixties, he was primarily concerned with physical reactions to the food additives and salicylates (e.g., asthma, hives, etc.). During the early seventies he became increasingly interested in H-LD because of the apparent magnitude of the problem and began exploring the possibility of a relationship between both food additives and salicylates and hyperactivity. He had noticed and received reports of behavioral effects of his diet (e.g., the much cited cases of his 1968 Oakland hives patient whose psychiatric symptoms were alleviated by adhering to the diet for two weeks), but had not devoted much time or energy to any follow-up. He reviewed this 16 type of cases in which there were behavioral responses to his diet and also looked at trends in the incidence of hyperactivity and increases in artificial additives in foods. This all eventually culminated in his hypothesis that by some unknown mechanism, salicylates and artificial food additives may cause H-DL in a genetically predisposed group of children (Feingold, 1973a, bi 1975a, b). In 1975b he wrote: "The most important and most dramatic adverse reactions induced by synthetic colors and flavors are behavioral disturbances. We know that any chemical compound, natural or synthetic, has the potential to produce an adverse response in a susceptible individual. We know that both drugs and food chemicals are low molecular (weight) compounds. We also know that drugs can influence the behavior of animals and men. Accordingly, it is reasonable to expect food chemicals to cause behavioral modifications. And with the thousands of food chemicals in the environment, it seems remarkable that their influence on behavior has not been suspected sooner" (p. 799). As mentioned earlier, Feingold (1973a, 1975a, b) admitted that the mechanism by which the adverse reactions are caused by ingestion of these substances is not known. He did feel, however, that the reactions in these genetically predisposed individuals appear to be pharmacologic, rather than immunologic in nature. He suggested that because the behaviors can be "turned on and turned off" by the diet and 17 there is rapid improvement once artificial flavors and colors are eliminated, "the pharmacologic action of the additives may serve as a repressor mechanism that prevents normal expression" and that "food chemicals induce a functional derangement rather than persistent organic changes" (1973a, p. 18). He (1975b) discussed some examples of interactions between heredity and drugs (e.g., Zurich hemoglobinopathy, isoniazid (INH) intolerance) which result in deleterious outcomes and suggested that these may possibly be models or parallel the mechanisms underlying the adverse reactions to food colors and flavors. Hannuksela nad Haahtela's (1987) review of literature on somatic reactions to common food addititives also suggested pharmacologic rather than immunologic action. Taylor (1979) felt that Feingold was describing a toxic rather than an allergic effect. In elaborating on this distinction he wrote: "A toxin causes damage to the body directly. An allergy, by contrast, is an alteration of the body's response to a substance. It is produced by the immunological defense mechanisms. It is not shown in the first encounter with the allergen, but only in later contacts. In the Feingold hypothesis, the toxic effect is supposed to be apparent only in predisposed individuals so the notion of an idiosyncratic response is still present" (p. 357). Swanson and Kinsbourne (1980) in their study assessing the effects of food dye challenges (to be 18 discussed later) also felt that the rather immediate effects and peaking pattern of dyes suggest a toxic or pharmacologic reaction rather than a sensitivity. On the other hand, Schneider (1945) in discussing possible causes of hyperactivity stated that it is possible that too little attention was being devoted to food allergy. He based his argument primarily on observations by Winkleman and Moore (1941) who reported nervous manifestations of allergies, Karnosh (1944) who felt that all levels of the nervous system are involved in allergic reactions, and Clarke's (1944) descriptions of children's reactions of headaches, unruliness, and inattention to wheat and chocolate milk. Similarly, Kittler and Baldwin (1970) believed that based on review of the literature and their own research and clinical experience, minimal brain dysfunction syndrome is caused by allergy of the tension­ fatigue syndrome classification. Feingold estimates, at different times, a favorable response rate between 30 and 50 percent (Feingold, 1975a, b; 1977). In 1975b he wrote: "The experience to date indicates that approximately 50 percent of children with H­ LD respond to strict elimination diets. Loss of hyperactivity, aggression, and impulsiveness are the initial changes observed. This is soon followed by improvement in muscular coordination as indicated by improved writing and drawing abilities, greater facility 19 with speech, and loss of clumsiness. Disturbances in cognition and perception are usually the last to respond. with an increased attention span which permits greater concentration, scholastic achievement improves rapidly. Age seems to be an important determining factor in the degree and speed of response to dietary management. The younger the child, the more rapid and more complete is the improvement. A 3 or 5-year-old child may show improvement as early as the third day of dietary control. Beyond this age period, two to three weeks of strict dietary management are usually necessary before a decided improvement occurs. In the adolescent, 15 years of age and older, the response is not only slower, but frequently less complete or shows little improvement. In the older child and adolescent, the motor disturbances, for example, impulsiveness, aggression, and hyperactivity, may be controlled with diet, but lack of improvement in cognition and perception interferes with learning abilities. This failure leads to frustration, withdrawal, and loss of self-esteem which may be followed by antisocial behavior. The highest percent of behavioral failures is seen in late childhood and adolescence" (p. 800). He also claimed.that while "50 percent have a likelihood of full response, 75 percent can be removed from drug management" (1975a, p. 71). He reported, too, that although response while on the diet may be gradual, "any infraction of the diet, either inadvertent or deliberate, 20 causes a recurrence of the complete behavioral pattern within two to four hours which persists for one to four days" (1975b, p. 801). It should be mentioned that Feingold modified his diet even further. As reported by Wender (1977), the current diet does not exclude foods previously eliminated because of supposed salicylate content. He gave two reasons for this revision. First, there is inadequate evidence that these foods do indeed contain measurable salicylate. Second, Feingold came to believe (communicated via personal correspondence with Wender) that salicylates are usually not the offenders. In 1977a, in a description of his diet, Feingold also included foods containing the antioxidant preservatives, butylated hydroxytoluene (BHT) and butylated hydroxyanisole (BRA) on the list of items to be avoided. In 1977b, he reported that the therapeutic response rate is probably even higher than the 30 to 50 percent estimate claimed earlier when foods containing these two compounds are eliminated. Studies on the Feingold hypothesis have yielded mixed results as pointed out by many (e.g., Conners, Goyette, Southwick, Lees, & Andrulonis, 1976; Harley, Ray, Tomasi, Eichman, Matthews, Chun, Cleeland, & Traisman, 1978; Prinz, Roberts, & Hantman, 1980; Williams, Cram, Tausig, & Webster, 1978; Rippere, 1983). There is still no conclusive evidence and more research is needed. Taylor 21 (1979) listed five problems associated with testing the hypothesis (also discussed in Conners, 1980; Cook & Woodhill, 1977; Lancet, 1979b). 1. Due to the nature of the diet and the degree of family involvement and effort required, there is great potential for placebo effects. Adequate controls are, therefore, imperative. 2. Since strict compliance is required, negative results have some time been attributed to "cheating." As is immediately evident when reviewing the diet, for various reasons, it is extremely difficult to follow. Lew (1977), a child psychiatrist, felt that she needed to experience the diet before prescribing it. She observed that "meals in general were far less appetizing in content and in appearance", "shopping time was prolonged", and "food preparation time was markedly lengthened" (p. 190). After 4 weeks she reported that her family expressed "great relief" at the termination of the diet and concluded that the diet is "very difficult to maintain in practice" and that "it would be impossible for a mobile child not to defect from it, unless perhaps the whole family partook of the diet" (p. 190). In addition, food labels are often incomplete. One of the reasons for this is that many food products are made from "secret formulas" which manufacturers guard carefully (Feingold, 1968). Wender (1986) also pointed out that Feingold was not specific 22 enough in regard to "which chemicals and to what concentrations should be allowed or excluded from the additive-free diet" (p. 37). 3. Feingold does not claim that all hyperactivity is attributable to food additives and salicylates. Therefore, they may be a sUbgroup of hyperactives who respond to the diet yet are overwhelmed in statistical analyses by a majority who do not. 4. Hyperactives are a very heterogeneous group and it is possible that very different groups of children are being tested in the studies. 5. As mentioned earlier, many regard food sensitivity and intolerance as belonging in the "realm of fantasy" and, therefore, not worthy of attention. There is another obvious consequence of Feingold's lack of specificity that has been frequently mentioned (e.g., sieben, 1977). Since the diet eliminates so many foods, it is difficult to identify the offenders. Furthermore, as stated in Lancet (1979b) a single synthetic flavor may include up to 50 different substances. Due to the pUblic response to the Feingold K-P diet, the Nutrition Foundation and the u.S. Department of Health, Education and Welfare formed study committees to review the results of research conducted on Feingold's hypothesis. The National Advisory Committee on Hyperkinesis and Food Additives (1975) reported to the Nutrition Foundation that 23 proponents' claims were impressionistic, anecdotal and lacking in objective evidence. The Committee also recommended guidelines for subsequent research. The Interagency Collaborative Group on Hyperkinesis (1975) comprised of representatives from all HEW agencies concerned with children, reported to the Secretary of Health that: "It is the opinion of the group that the studies have neither proven nor disproven the hypothesis that a diet free from artificial colors and flavors reduces the symptoms in a significant number of children with the hyperkinetic behavior syndrome. The group feels, however, that the evidence taken as a whole is sufficient to merit further investigation into the relationship of diet and the hyperkinetic syndrome." This Committee also recommended specific research procedures for future studies. Many articles on the Feingold hypothesis have been pUblished in the last decade. Unfortunately, as pointed out in the Nutrition Foundation report, most have been clinical observations and anecdotal accounts. There have been less than 15 studies that have met minimal standards of appropriate research design (Kavale and Forness, 1983; Wender, 1986). The most often cited studies on the diet or aspects of the diet can be grouped into (1) uncontrolled reports, (2) controlled comparisons between the Feingold K­ P diet and a control diet, and (3) challenge studies. 24 As is obvious, uncontrolled reports are of little value because of the failure to control for possible extraneous sources of variance. Salzman (1976) put 15 children who had responded positively when tested for sensitivity to salicylates, artificial colorings and flavors by the Hawley-Buckey (1974) method on the Australian Version of the Feingold K-P diet. They reported that within 4 weeks, mothers' ratings showed improvement in 14 of the 15 children. In another Australian study, Cook and Woodhill (1976) reported that the parents of 10 children of their sample of 15, were "quite certain" and those of an additional 3 were "fairly certain" that their children's behavior improved noticeably with the diet and also deteriorated promptly when there were significant deviations from the diet. As Werry (1976) and others (e.g., Harley et al., 1978) have cautioned, these studies contain many methodological weaknesses. For example, compliance to the diet was apparently not monitored, there were no matched control groups or crossover designs, blind procedures were not employed, and both failed to use standardized, validated instruments. There ar~, therefore, many plausible alternative explanations for the reported favorable responses to the diet. In another study, Brenner (1977) reported "marked improvement" in 11 of 32 children. Although results of a 25 "control group" were also reported, the comparability of the groups are questionable. The criticisms of the two Australian studies also apply (i.e., no control for placebo effects, compliance not monitored, and objective outcome measures were not used). stine (1976) worked with 2 hyperactive children who presented overwhelming problems in school and at home and who experienced unpleasant side-effects to their medication regimens. As a last resort, these children were put on the Feingold K-P diet and dramatic alleviation of the most troublesome symptoms were reported in 10 weeks for one child and 5 months for the other. The author suggested that a more sudden and drastic improvement may not have been realized because of concomitant severe cognitive and emotional problems in both children. The design weaknesses discussed above also apply to this study. Palmer, Rapoport, and Quinn (1975) attempted to test Feingold's claim that hyperactive children consume larger amounts of food additives than normal. Theyobtained samples of the diets of 56 hyperactive boys and a control group of 23 boys who showed no learning or behavior problems via dietary questionnaires completed by the parents of these children. They found no significant differences. However, the reliability of the parents' responses and the validity of the method utilized for evaluating their responses is questionable. 26 To date, there have been three controlled studies that have compared the Feingold K-P diet with other diets. Conners et ale (1976) conducted a double-blind crossover trial with 15 hyperactive children, ages 6 to 12 years, 11 months. Nine of the children received the Feingold K-P diet first and the others received a control diet first. Parents and teachers rated the children's behavior weekly using Conners Questionnaires (Conners, 1969, 1970). The teacher questionnaire consists of a list of behavioral problems commonly encountered in school-age children (e.g., constantly fidgeting, hums or makes other odd noises, disturbs other children). The teacher rates each item on a 4-point scale to reflect the degree to which each problem is exhibited by a particular child. The parent questionnaire consists of a lengthier list of behavioral problems (e.g., will not eat enough, cannot fall asleep, cannot be left alone, always climbing). The parent indicates on a 4-point scale the degree to which their child is bothered by each problem. Most of the analyses in this study were focused on a 10-item "hyperkinesis index" subscale which measures the primary sYmptoms of hyperactivity. The ten items are identical for both questionnaires. The teachers reported significant reductions of hyperactivg symptoms with the Feingold K-P diet as opposed to the control diet but the parents did not. The authors 27 suggested that it may be that effects are more noticeable in school where there is more structure and greater demands on children's attention and goal-oriented behavior. The ratings for individual children did show, however, that "four or five of the children" were rated as improved by both parents and teachers. Based on this finding, the authors suggested that a small subgroup of hyperactives may respond to this diet. It should be mentioned, too, that although parents and teachers differed in their comparisons of the Feingold K-P and control diets, both found behavior on the Feingold K-P diet to be significantly improved over baseline and also 'not significantly different between baseline and the control diet. In addition to the differences between teachers' and parents' ratings, a rather marked order by treatment effect also emerged to weaken the argument for the diet (authors, however, claimed this interaction was not significant). Most of the favorable responses to the diet occurred when it followed the control diet. The authors pointed out that this is quite common in psychopharmacologic research which use crossover designs. They cited as an example a study conducted by Conners, Eisenberg, and Barcai (1967) on the effects of dextroamphetamine on hyperactivity in which teacher ratings reflected a much more pronounced drug effect in a group that received a placebo first. Rapp (1979a) suggested that "when the Feingold-like diet was 28 given first, the beneficial effects extended into the placebo diet period, obscuring proper evaluation of the diets, but that when the placebo diet was given first, the true effect of each diet was more obvious" (p. 10). As previously mentioned, however, Feingold did claim that dietary infractions resulted in rather immediate recurrence of hyperactive symptoms. In addition to these concerns, many have expressed other reservations about this study. Taylor (1979) indicated that the two diets may not have been adequately matched and, therefore, the mothers not truly blind. Levine and Liden (1976) and Wender (1977, 1986) expressed similar concerns. Taylor also questioned whether the effects observed were clinically significant, for 10 of the 15 sUbjects were rated in the clinician's general ratings as minimally changed, unchanged, or worse. Along similar lines, Harley et ale (1978) pointed out how even the improved teachers' mean rating of 13.93 while on the diet was very close to the cut-off score of 15 used in the study

as ~n indicator of hyperactivity. other criticisms included the measurement drift that has been associated with the Conners' Rating Scales (Werry, Sprague, & Cohen, 1975)--they found that there was an improvement in scores with repeated testing even if no treatment was administered, minimal precautions against breach of diet (Levine and Liden, 1976), small sample size 29 (Mattes, 1983; Taylor, 1979), lack of other objective laboratory and observational measures (Wender, 1986)~ and inadequate description of the sample (Levine & Liden, 1976). Dickerson and Pepler (1980), Mayron(1979), and Rapp (1978a) proposed that the ambiguous results may have been due to sUbjects being sensitive to and reacting to some foods permitted in the Feingold K-P diet. Mayron, therefore, felt that in order to reliably test dyes, flavors, and salicylates, the children's other food and chemical sensitivities must first be identified. Rapp (1978a) with this concern in mind, designed a study to test the effects of dyes (and some particular foods) on the sample of 5 to 16-year-old hyperactive children. She first tested each child with food dyes and six different foods using a SUblingual technique (i.e., allergenic extracts are placed under the tongue and symptoms in sensitive individuals supposedly occur rapidly). Measures included coding and motor accuracy tests and observations of activity level by a nurse and housewife. She reported marked or moderate increases in activity in 9 of the 24 children in response to the dye tests and similar increases in 6 of the 24 children in response to the food tests. There were no consistent changes on the other measures. The children were then put on a diet that excluded the six foods (i.e., milk, wheat, 30 egg, cocoa, corn, and sugar) and synthetic colors for 7 days. Parents reported moderate to marked improvement in activity for 12 of the 23 children. Abbott Hyperkinesis Index Sheets mean scores for these 12 children were also sUbstantially lower than the mean scores of 11 children who had not been reported as improved. After the 7-day diet, each of the excluded foods and foods containing dyes were reintroduced, one each day, and the reactions noted. Eleven children were reported to have reacted with hyperactivity when dyed foods were eaten and 10 of 21 children experienced some reaction to both food coloring and at least one food. Similarly, Egger, Carter, Graham, Gumley, and Soothill (1985) reported that everyone of the 76 hyperactive children (2 to 15-year-olds) studied had more than one food "sensitivity." For all children, significant improvements in response to a synthetic food dye and preservative-free diet only occurred when at least one other food was also eliminated. The foods that most frequently needed to be eliminated included milk, chocolate, wheat, egg, corn, and cheese. There are some critical procedural weaknesses associated with Rapp's study that need to be recognized. First, the diet phase was not conducted blind. Second, the children may have been able to distinguish between the dye and placebo (grape juice), and food test material and 31 placebo (prune juice). Third, some of the children were on medication. Fourth, dietary infractions were reported. Harley et ale (1978) also conducted a double-blind crossover study comparing the effects of the Feingold K-P diet to a control diet in 36 school-age and 10 pre-school hyperactives. The study was conducted during the spring, summer, and fall. The spring and summer groups were on each diet for 3 weeks, the fall group for 4 weeks. Diets were assigned in random order. Elaborate measures were taken to ensure dietary compliance. For example, all previously purchased foods were removed from the participants' homes and each family's entire food supply was delivered to them weekly, all family members participated, and supplementary food was provided for special occasions such as holidays and school parties. other procedures were also included to obscure the dietary manipulations. For example, there were special production runs to produce identically packed chocolate bars and cakes (some containing additives and others additive-free), innocuous food items were systematically introduced and withdrawn, and items such as potato chips were introduced and withdrawn to give the impression of diet changes when in fact, depending on the brand, the item was permitted in both diets.

An average of three classroom observations were obtained weekly throughout the study. Behaviors observed 32 were attending-to-task, restless motor activity, and classroom disruption. The Conners Parent and Teacher Questionnaires were completed weekly. In addition, at the end of each diet phase, neuropsychological data (e.g., measures of motor speed and coordination, reaction time, concentration/attention) and laboratory observations (i.e., locomotor activity, attending-to-task under free play and restricted-activity, instructional conditions) were obtained. The overall results did not provide convincing support for the efficacy of Feingold K-P diet. For the school-age sample, favorable ·assessments of diet effects were most frequent in the sUbjective parent ratings. There were very little positive effects noted by the teachers and virtually no evidence of effectiveness from the neuropsychological, classroom, and laboratory observational data. Similar to the Conners et ale (1976) study, a significant order effect was evident in the parent ratings with favorable responses associated with the diet only when it followed the control diet. For the pre-school sample, 10 mothers and 4 of 7 fathers rated their children's behavior as improved on the diet and there were no diet by order effects. The authors, however, chose not to emphasize these findings because: (1) the sample consisted of 10 as compared to 36 in the school­ age group, (2) no teacher ratings were obtained (children 33 were not old enough to attend school), (3) there were no corresponding changes in the neuropsychological test ,scores and the laboratory observational measures, (4) the Conners Questionnaire was used as one criterion for selection despite the fact that it was developed and validated on school-age children and more generally, (5) it is difficult to develop reliable and valid criteria for diagnosis and measurement of hyperactivity in this age group. Some of the authors concluded, however, that although there is much evidence against the Feingold hypothesis for school-age children, it is not possible to dismiss the possibility that hyperactive symptoms in pre-school age children may result from synthetic colors and flavors (Harley, Matthews, & Eichman, 1978). Although this study is acknowledged as the best evaluation of the Feingold K-P diet (e.g., Bock, 1986; Kolata, 1978; Taylor, 1979), it has not escaped criticism. Werry (1976) felt that the only weakness in the design was the policing of compliance. Wender (1986), despite the painstaking measures taken to insure blind conditions, still felt that an "informed family" could easily distinguish between the two diets. As reported by Kolata (1978), Feingold responded to the conclusions of this study by claiming: (1) the results were biased because the study was funded by the partisan International Life Sciences Institute-Nutrition Foundation which is sponsored and 34 directed by large conglomerates within the food and drug industry including General Foods Corporation, Nestle Food corporation, the Coca-Cola Company, International Flavors and Fragrances, Inc., Abbott Laboratories, and Lilly Research Laboratories, (2) the children deviated from the diet when they were in school, and (3) 4 of the 36 school­ age children were rated as improved by teachers and parents but were overshadowed by the majority with contradicting parent-teacher assessments. In a much less rigorous but more recent study, Gross, Tofanelli, Butzirus, and Snodgrass (1987) tested 39 learning disabled children during a 2 week summer camp. Eighteen of these 11 to 17-year-olds were also diagnosed as hyperactive with 17 taking the usual medications. The youngsters received the Feingold K-P diet during the first week and a diet "with as much artificial coloring and flavoring as possible" (i.e., included soft drinks, pastries, candies, etc.) during the second. The youngsters were not told about the study and according to the authors, the Feingold K-P diet was "meticulously observed". The dining room was scanned by a television camera for 4 minutes at each mealtime. These tapes, excluding those filmed on the first and last days of the camp, were viewed in random order and rated by the senior author and two teachers who had not been at the camp. A 0 to 10 scale was 35 used to rate the group on the degree of motor restlessness, disorganized behavior, and misbehavior exhibited. The ratings revealed no differences between the 2 weeks. Based on this finding, the authors concluded that the Feingold K-P diet does not appear to benefit most children with learning disabilities or hyperactive children who are being effectively treated with medication. However, because two children had to be sent home during the second week due to misbehavior and the camp director and teachers (who were not blind) did feel that the children were slightly noisier and more active during the second week, the authors suggested further study. The design of this study is obviously weak in a number of areas. For example, a) the measures used appear to have been inadequately sensitive (i.e., rates were under one in both weeks), b) observations in different settings would have been desirable, c) only overall group ratings were obtained, d) 4 minutes of observation during mealtimes are insufficient, e) a week is a short trial of a diet, f) the diet during the second week was not well-defined, and g) the hyperactive children were taking their medication. In short, this study was seriously flawed and the results and conclusions are highly questionable. The third group of studies used challenge designs. In a follow-up study of Conners et ale (1976), Goyette, Conners, Petti, and curtis (1978) tested the effects of 36 artificial colors on 16 children (mean age 8.3 years) who showed at least a 25 percent improvement during a month long, nonblind trial of the Feingold K-P diet. These children were kept on the diet and in double-blind fashion, given alternating (ABAB or BABA) challenge or placebo substances in chocolate cookies. Each child received two placebo or two challenge cookies containing a total of 26 mg of food coloring per day in 2-week alternating sequences for 8 weeks. There were no statistically significant effects. There was, however, evidence of transient deterioration in performance in the younger children on a visual-motor tracking task. A younger group (mean age 6 years) who had also shown improvement on the Feingold K-P diet was then studied and a statistically significant increase in hyperactive behavior as observed by parents was found when ratings were taken within 3 hours of the challenge. The authors concluded that the data suggested that synthetic colors impair and disrupt the behavior of children and that younger children may be more likely to be significantly affected. As pointed out by.Taylor (1979), these results must be interpreted cautiously, for the sample was very small, improvement was shown on a sUbjective rating scale which was used in a manner for which it wasn't designed or validated, and the active and placebo cookies may not have 37 been indistinguishable. In addition, "chocolate is commonly known to be a potent and prevalent food allergen in its own right" (Rippere, 1983, p.20), and that sUbjects' inability to tolerate chocolate cookies was noted by Conners (Wender, 1977) and Williams, Cram, Tausig, and Webster (1978). In two follow-up studies to Goyette et al. (1978), Conners (1980) and Conners, Goyette, and Newman (1980) found no significant effects of food color challenges on measures of hyperactive behavior, learning, attention, and activity. Harley, Matthews, and Eichman (1978) conducted a follow-up of Harley et al. (1978) also using a double­ blind, multiple crossover design. Nine hyperactive boys (mean age 9 years, 3 months) who were selected because of favorable responses to the Feingold K-P diet in the initial study were kept on the diet and administered alternating challenge and placebo trials. The measures were similar to those utilized in the earlier study, that is, parent and teacher ratings, observations of classroom behavior, and neuropsychological test scores. Also, as in the previous study, all food was provided to the families. Cookies and candy bars were used as challenge and placebo foods with 27 mg/day of food colors being administered during challenge phase. No statistically significant results were obtained on any of the measures. Only one child exhibited a pattern of 38 parental ratings and classroom behavior that approximated the predicted responses for the challenge and placebo phases. Some of the criticisms voiced by Feingold and Wender (1976) may also apply to this second study. Concerns about the small sample size and the issue of possible sensitivity to chocolate also need to be considered. Williams, Cram, Tausig, and Webster (1978) compared the effects of stimulant medication and a diet free of artificial colors and flavors in 26 6- to 14-year-old hyperactive children. The authors created four conditions: stimulant medication-cookies with food coloring, stimulant medication-placebo cookies, placebo medication-cookies with food coloring, and placebo medication-placebo cookies. Each child was put on the modified Feingold K-P diet and spent a week in each condition. The children ate one cookie in the morning and another in the afternoon in all conditions. During the active cookie phases, each cookie contained 17.6 mg of dyes (35.2 mg/day). Conners Questionnaires were completed every day by parents and every other school day by teachers. Double-blind conditions were maintained throughout the study. The results showed that stimulant drugs (methyl­ phenidate or dextroamphetamine) produced significant improvements in behavior as rated by both parents and teachers regardless of diet status (i.e., type of cookie). 39 statistically significant favorable diet effects were only found with teachers' ratings when the children were receiving placebo medication, that is, when receiving the placebo, the children displayed more hyperactive behavior in school after eating the active cookies. There were clearly significant improvements related to the diet in about 1/4 of the children. The authors concluded that the medications were definitely more effective than the diet in reducing hyperactive behavior and that the data also suggested that their modified diet may reduce hyperactive behaviors in some children. In the discussion, they mentioned some factors that are important in interpreting the results: (1) dietary infractions occurred, (2) behavioral checklists were the only outcome measures used, (3) the children were in each condition for only 1 week so there's no indication of what any long-term effects may be, and (4) the sample of children was probably heterogeneous. In addition, Taylor (1979) felt that the interaction of treatments was confusing. That is, for parents, there were favorable diet effects when the children were also receiving medications and for teachers, the opposite was true. Levy, Dumbrell, Hobbes, Ryan, Wilton, and Woodhill (1979) investigated the effects of a single food coloring, tartrazine, on 22 hyperactive children between the ages of 4 and 8 years. The 5 mg/day tartrazine challenge, while on 40 the Australian version of the Feingold K-P diet, did not result in any overall effects. Using the criterion for sUbject selection that Goyette et al. (1978) utilized in their study, they selected a subsample of 13 subjects who experienced a 12% to 15% reduction of hyperactive symptoms on the diet and conducted post hoc analyses. Mothers' ratings for this subsample during the double-blind challenge and placebo trials showed significantly more hyperactive behavior during the challenge periods. However, as the authors admitted, the biscuits containing the tartrazine were not always identical to the placebo biscuits and since the objective tests (e.g., Sprague Ballistrographic Chair, Continuous Performance Test, Draw­ A-Line Slowly Test, an auditory sequential memory task, etc.) and teachers' ratings did not also reflect the challenge effects, the mothers' ratings may have been due to a placebo effect. In a sUbsequent replication study (Levy & Hobbes, 1978), 8 hyperactive children (mean age 5 years, 2 months) who were selected by the Goyette et al. (1978) criterion were given challenge or placebo cookies randomly over 14 days. On challenge days, 4 mg doses of tartrazine were administered via the cookies on challenge days. Although the mothers' scores were higher on challenge days, the effect was not significant. The authors speculated that failure to obtain a significant effect may have been due to 41 small sample size, the inadequacy of the Conners Questionnaire when differences are small (i.e., too insensitive), and failure to use more objective measures. Rose (1978) also examined the effects of tartrazine by studying two ll-year-old girls using a variant of the BAB design with a double-blind format. Both girls had been on the Feingold K-P diet for at least 11 months prior to this study. In the treatment phases, each child was given an

"oatmeal-type" cookie containing 1. 2 mg of tartrazine at breakfast each day. A cookie without the dye was given each day during the placebo phases. These phases were repeated once for each subject. The dependent variables were: on-task behavior (duration), out-of-seat behavior (frequency and duration), and physical aggression (frequency). Data were collected in school by trained graduate students. The results showed that for both sUbjects, there were increases in frequency and duration of out-of-seat behavior and decreases in duration of on-task behavior during challenge phases. The frequencies of physical aggression showed no definite pattern possibly because of the extremely low rates th~oughout the study. Several problems were encountered in this study. First, because one of the children attended a private school which frequently cancelled classes to give their students opportunities to participate in enrichment activities, 42 there were fewer observation periods and challenge phases "occasionally" occurred before a steady-state condition was achieved in nonchallenge phases. Second, dietary infractions were reported for both sUbjects. Third, because of a "variety of externally imposed conditions" there were no in situ reliability checks. David (1987) investigated the effects of this dye, too, by challenging 24 boys and girls between the ages of 1.6 and 12.4 years whose parents had reported histories of purely behavioral reactions to this additive. In addition, since the parents of six of these children also reported similar adverse behavioral changes to benzoic acid, all were also challenged separately with this preservative. According to the author, most of these children were labeled as hyperactive although only six actually fulfilled the DSM-III criteria for this disorder. All children were on diets that avoided at least two chemicals at the time of the study (many of the sUbjects were also restricted from eating various other foods). All parents reported that previous dietary infractions resulted in observable adverse behavioral reactions within 2 hours. The double-blind procedures were conducted in a pediatric ward. On the day of the challenge, the children were given 50 mg of tartrazine in a fruit drink that they were not sensitive to and a second 250 mg dose at least 2 hours later. The benzoic acid challenges followed on 43 another day via the same procedures and with the same doses. At least one parent and the nursing staff observed the children in free play conditions and were instructed to keep individual reports of any behavioral changes. There were no behavioral effects reported by parents and nurses in either the active or placebo conditions. As a result, the parents of 21 of the 24 sUbjects were convinced to return their children to a normal diet. It was further reported that at an unspecified follow-up, all children were free of food-related problems. However, as the authors pointed out, there were some significant procedural weaknesses in this study. First, four children were withdrawn before being challenged with benzoic acid because their parents felt that it was unnecessary to continue after no reactions were observed in the tartrazine challenges. Second, this study was conducted in a hospital which was at best unfamiliar and at worst a very frightening environment. The children's behavior may have been influenced by this setting. It is specifically noted in the DSM-III that hyperactive symptoms may not be displayed in new situations. Third, a valid and reliable scoring system was not used. Parents and "non­ psychiatrically trained nursing staff" were merely asked to record any changes in behavior that they observed. The necessity of objective measures cannot be overstated. Fourth, as in many challenge studies, all children had been 44 on diets that eliminated foods that their parents felt they were sensitive to and they may have, therefore, been able to recover from previous repeated or constant exposures and been less apt to react. Finally, as suggested by some (e.g., Randolph & Moss, 1981; Rapp, 1978a; Tryphonas & Trites, 1979) reactions may occur only as a result of additive exposures or because of interactions between agents. These and other general issues will be discussed more fully later. It would have been interesting to see if symptoms appeared when the children were also exposed to the other offending foods in systematic and controlled procedures. Due to these weaknesses and questions alone, the results cannot be taken as firm evidence and the author is correct in concluding this study does not preclude the possibility of purely behavioral reactions to these additives. In a more recent study, Rowe (1988) tested nine "suspected hyperactives" with this dye and another azo dye, carmoisine (red). Prior to the challenges, these children had participated in an open trial of the Feingold diet and displayed 6 months of stable, improved behavior. While maintaining the additive-free diet, the children were tested over 18 weeks with the dyes and a placebo (lactose). The children took capsules. each morning containing the placebo or 50 mg of either tartrazine or carmoisine. In double-blind format, the dyes were each tested for one week 45 on two separate occasions for a total of 4 weeks of dye challenges. There were 2 or 3 week washout (placebo) phases between each active phase. Checklists of eight behaviors often associated with dye ingestion were completed by parents daily and by teachers when possible. They simply indicated for each behavior whether it was a "good day" or a "bad day". Two children reacted significantly to both food colorings. As rated by her parents, a 7-year-old demonstrated increased activity, irritability, low frustration tolerance, sleeplessness, and short attention span. Similarly, the other reactor, a 12-year-old boy, displayed increased activity, low frustration tolerance, aggression, short attention span, and sleep disturbance. The author concluded that the results obtained on these two subjects demonstrated "a significant dependent relationship between the ingestion of azo dye food colorings (tartrazine and carmoisine) and behavioral change". Just as importantly, he also discussed possible

rea~ons for the equivocal results of many of the earlier dye studies. He proposed that the selection of subjects based on DSM criteria may be inappropriate for common behavioral responses to dyes are irritability, restlessness, and sleep disturbance rather than those associated with attention deficits. These studies may have excluded potential reactors because of their method of 46 sUbject selection. He also suggested that the dye doses were inadequate and the Conners scale is inappropriate (both issues to be discussed later) and urged the development of a valid and reliable behavior rating instrument for dye studies. One serious weakness of this study as acknowledged by the author was the reliance on such a simplistic rating scale. It was obviously very limited in scope (range of behaviors) and the crude scoring format of dichotomous categories did not allow fine discriminations in the persistence or severity of each behavior. Secondly, all teacher checklists were imcomplete and there were low correlations between parent and teacher ratings. Thirdy, the washout period between the active challenge weeks was not long enough for the male reactor's behavior to return to baseline levels. It took 3.5 weeks for his behavior to return to pre-study levels after the last active challenge and the placebo phases were only 2 to 3 weeks in length. The effects of one active phase likely ran into the next. Finally, both reactors committed quite a few dietary infractions during the 18 weeks (i.e., 8 for the girl and 7 for the boy). Mattes and Gittelman-Klein (1978) conducted a double­ blind mUltiple crossover study with a 10-year-old hyperactive boy whose behavior, according to his parents, definitely improved on the Feingold K-P diet. The child 47 was kept on this diet and 11 one-week independent trials in which the child was given either placebo cookies of all natural ingredients or active cookies containing food colorings (each contained 1/5 of the average daily intake of food colorings in the United states) were run. His mother and teacher used the Conners Questionnaires to rate his behavior during the trials. The mother and teacher ratings were unaffected by the conditions. However, the mother correctly guessed, beyond chance levels, which type of cookie her son had consumed in particular trials. The mother's guesses were based on the degree of irritability that was displayed by her son rather than the typical hyperactive symptomoloqy. The question­ naire does not measure irritability. In a subsequent study, Mattes and Gittelman (1981) studied 11 children, ages 4 to 13 years, who according to a Feingold Association, responded very favorably to the diet and reacted markedly to synthetic coloring. While on the diet, the children were challenged in a double-blind format for a week with the doses of the food dyes increased daily. A 13 mglday dose contained in a cookie was administered on the first day and incrementally increased to 78 mg/day on the 6th and 7th days. No significant differences were obtained between the placebo and challenge phases on ratings by teachers, parents, and a psychiatrist nor on a test of distractibility. They discussed several possible 48 reasons for the negative findings: (1) active and placebo cookies may not have been accurately labeled (a concern of the Feingold Association), (2) most of the samples did not meet the usual criteria for hyperactivity, and (3) small sample size. Spring, Vermeersch, Blunden, and Sterling (1981) conducted a study in which 6 hyperactive boys were challenged with synthetic colors in a single-subject, double-blind, double crossover format. These children (ages 8 to 13 years) had been on the Feingold K-P diet before the study was organized and had, according to their mothers, displayed less hyperactive symptoms. There was a 2 week baseline period and a 6 week experimental phase. The subjects were challenged with 26 mg/day doses of food coloring concealed in chocolate cookies during 2 nonconsecutive weeks. They were given placebo cookies during the three weeks. An abbreviated lo-item hyperactivity rating scale (almost identical to Conners Questionnaires) and guesses from the mothers and teachers as to the treatment and placebo weeks were used as outcome measures. A significant positive relationship between the challenge and mothers' and teachers' guesses and scale ratings was found for only one of the children. However, a subsequent 5-week follow-up on this child failed to replicate these results. 49 There were several factors that may have contributed to these findings. First, the mothers and teachers of all the children reported dietary infractions with reports of more than 1 per week for 3 of them. These, however, were only what was reported. Before the study, these 8 to 13­ year-olds signed agreements to strictly follow the diet during the study. Second, chocolate cookies were used in the challenge and placebo trials. Third, objective, validated measures were not utilized. Swanson and Kinsbourne (1980) suggested that lack of responses to dye challenges may be due to insufficient doses of dye. They therefore conducted a challenge study in which 100 and 150 mg/day doses of a blend of nine food colorings were administered to a matched group of hyperactive and nonhyperactive children (mean age, 10 years). The hyperactive group was composed of children who responded favorably to stimulant medication and who showed other symptoms of hyperactivity. The nonhyperactive group was comprised of children who reacted adversely to stimulant medications. The subjects were admitted to a hospital and maintained on a diet free of artificial dyes and additives for 3 days. On days 4 and 5 the children received either active or placebo (sugar) capsules. The order of challenge and placebo capsules were counterbalanced across the subjects. Ten children from each group (hyperactive and nonhyperactive) were given 100 50 mg doses and the other 10 in each group were given 150 mg doses. On each of the days, paired-associated learning tests were administered: 1/2 an hour before the capsules were taken and then at 1/2, 1-1/2, and 3-1/2 hours after the capsules were taken. The Conners Questionnaire was also completed twice daily by observers. No differences between the active and placebo trials were obtained on the questionnaire scores. There were, however, significant differences in performance on the learning tests between the two different conditions for the hyperactive group but not for the nonhyperactive group. The interaction between the time of testing and challenge was also significant with the effects of the dyes taking over 1/2 hour to become evident, peaking at 1-1/2 hours, and lasting for at least 3-1/2 hurs. There were no significant effects related to dose (100 mg vs. 150 mg) indicating that even the lower dose was sufficient to affect performance on the learning tests. Wender (1980) discussed several issues that should be considered in interpreting these results. First, the designation of hyperactive and nonhyperactive may be misleading. The mean score for the nonhyperactive group on the Conners Teacher Questionnaire was 12.3 with scores between 0 and 9 being within the normal range and a score of 15 being 2 standard deviations above the mean (usual criterion for a diagnosis of hyperactivity). Second, the 51 learning tests were repeated over a short period of time thereby possibly reflecting practice and/or fatigue effects. Third (related to the second), the different placebo response patterns of the two groups (i.e., decrease in errors over time for the hyperactive group and increase in errors over time for the nonhyperactive group), hinders interpretation of the effects. Mattes (1983) concurred with this last point and added that analysis of order effects which are necessary in evaluating studies utilizing crossover designs were not reported and that identifying sUbjects as hyperactive based on their reactions to stimulant medication is idiosyncratic. Weiss, Williams, Margen, Abrams, Caan, Citron, Cox, McKibben, agar, and Schultz (1980) studied 22 children (between 2.5 to 7 years of age) who exhibited behavioral problems and who had responded favorably to a diet free of synthetic colors and flavors. In a double-blind, mUltiple crossover design, each sUbject (while on a diet similar to the Feingold K-P) was given a 35.2 mg/day dose of a blend of seven food colors contained in a cranberry-colored soft drink. The challenges occurred on 8 days distributed randomly during the 77~day study. A placebo soft drink of caramel and cranberry coloring was served on the other days. Parents' ratings on symptoms that each had identified as being associated with their child's dietary infractions and scores on an abbreviated 10-item Conners 52 Parent Question-naire were used to evaluate the effects of the food colors. The Conners Questionnaire was completed once daily and the other ratings were obtained several times each day after the drink was taken. Twenty of the 22 children displayed no adverse reactions to the challenges. However, one three-year-old boy showed marked increases in two target behaviors as rated by his parents. The behavior of another child, a 34­ month-old girl, deteriorated significantly when challenged. Her mother was able to correctly identify five of the six challenge days during the 77-day study. Mattes (1983) cautioned that the statistically significant effects for only this one child could be a chance finding. Wender (1980) pointed out that changes were reported by parents in other studies (i.e., Mattes & Gittelmen-Klein, 1978; Harley, Matthews, & Eichman, 1978) but were not corroborated by other measures. As briefly described earlier, Egger, Carter, Graham, Gumley, and Soothill (1985) investigated the effects of food colorings, preservatives, and many different foods on a sample of 76 children (ages ranging from 2 to 15 years) who were either assessed as being hyperactive or behaviorally disturbed with overactivity as a prominent feature. This study was conducted in three phases. In phase one, all children were put on an "oligoantigenic" diet for 4 weeks. This diet typically consisted of two 53 meats, two carbohydrate sources, two fruits, vegetables, water, calcium, and vitamins. In phase two, those who had shown improvement were given foods not included in the diet. A different food was reintroduced each week with sUbjects eating at least one full serving of that food daily. If reactions occurred, the food was withdrawn, if not, the food was incorporated into the diet. Along with the various foods, food coloring (tartrazine - 150 mg/day for 1 week) and a preservative (benzoic acid - 150 mg/day for 1 week) were also served in fruit juice. The fruit from which the juice was derived and sugar were also tested. In the final phase, those for whom a neutral diet had been identified and who had shown at least one reaction, entered a "double-blend, crossover, placebo­ controlled trial of reintroduction of one incriminated food." The foods tested included cow's milk, orange juice, wheat, capsules of tartrazine, and capsules of benzoic acid. The dependent measures included scores from Conners Parent Questionnaire, another questionnaire of symptoms completed by parents, behavior rating scale completed by one of the authors, matching-familiar-figures tests, Porteus Mazes, and actometer readings. Parents' ratings showed that 62 of the 76 children improved on the oligoantigenic diet with 21 of these children moving into the normal activity range. According 54 to the scores on the Conners Questionnaire, behavior was worse during the active (A) challenges than the placebo (P) periods. There was, however, a significant order effect. Those receiving the placebo first were given higher scores for the placebo period than those who received it last. Although the mean score was lower for the placebo condition in the PA group than the AP group, the difference was not significant. There were no significant differences on the other measures. However, except for the actometer readings, all other scores favored the placebo periods. Of interest were the findings that the most common offenders were benzoic acid and tartrazine and that no sUbject reacted to these alone. Forty-six other foods also caused reactions. The authors suggested that the lack of significant effects on most measures may be because they were inappropriate or not sensitive enough. However, in addition to the order effects, small N (authors admitted that most measures could be administered only to subgroups of the children), and comparability of placebo and active materials also need to be considered. Several authors (Harley & Matthews, 1979; Lipton & Mayo, 1983; Mattes, 1983; Wender, 1977, 1986) have reviewed the numerous Feingold K-P diet studies and attempted to summarize the results. All of them felt that the experimental data provided little evidence of dietary 55 effects on hyperactive symptoms. Wender (1986) reviewed 13 of the more well designed studies (all of which have been described above) and concluded that of the approximately

240 children evaluated in these studies "only 1% demonstrated any consistent behavioral change in the expected direction (p. 41)." Similarly, Harley and

Matthews (1979) wrote: "It must be emphasized that when data are collected via rigorous experimental designs which minimize potential sources of sUbjective bias and positive behavioral outcome expectancy, diet-related behavioral changes are much less spectacular than are the impressive clinical case studies or fervent parental testimonials which have been so widely disseminated (p. 129)." They also wrote as did others (e.g., Conners et al., 1976;

Goyette et al., 1978; Harley et al., 1978; Levy et al.,

1978; Taylor, 1984; Weiss, 1982; Williams et al., 1978) that there are, however, some data that provide evidence that a subgroup of children, especially younger children, may respond favorably to the diet. Due to this apparent sub9roup of responders and the various methodological issues that were discussed, no firm conclusions can be reached at this time. Although some have "closed the book," so to speak, on this type of research and see no need for further studies (e.g., National Advisory Committee on Hyperkinesis and Food Additives, 1980), many others

(e.g., Conners et al., 1976; Consensus Development 56 Conference, 1982; Levine & Liden, 1976; stare, Whelan, & Sheridan, 1980; Taylor, 1979; Thorley, 1983; Weiss, 1986; Williams et al., 1978) express the need for further research. Kovale and Forness (1983) claimed that the reviews of the empirical evidence on the Feingold hypothesis (e.g., Harley & Matthews, 1980; Sheridan & Meister, 1982; Stare, Whelan, & Sheridan, 1980; Tryphonas, 1979) have been "primarily narrative integrations resulting in impressionistic and sUbjective conclusions falling short of rigorous scientific standards for accumulating evidence" (p. 324). They, therefore, applied Glass' (1976, 1977) method of meta-analysis as "a means of statistically integrating a body of literature and providing a rigorous alternative to the typical narrative discussions of research studies" (p. 325). In this procedure, the treatment effects of studies investigating the relationship between the Feingold K-P diet and various outcome measures were converted into a common metric, an effect size (ES) statistic, that was independent of statistical significance. For each treatment effect, an ES statistic was calculated by sUbtracting the mean of the control group divided by the standard deviation of this group from the mean of the treatment group. A positive ES statistic was interpreted as indicating support for the Feingold K-P diet and a 57 negative statistic as indicating that the diet was ineffective. The authors felt the value of the ES statistic is that it allows "for a full statistical integration of a collection of studies which includes not only a description of the empirical findings on a common scale but also a description of how findings vary from study to study" (p , 325). ES statistics were calculated for 23 studies (total of 125 ES measurements), most of which have been reviewed above. It was found that, in toto, these measures did not reflect any significant favorable effects of the Feingold K-P diet. However, a small treatment effect was indicated. That is, sUbjects who experienced the diet were better off than 55% of the control sUbjects. The results of their analyses were, therefore, similar to the conclusions of the review articles discussed earlier. It is important to note that although some of the studies included in the analyses did test the Feingold K-P diet in its entirety, many only tested particular substances (i.e., primarily artificial food coloring). Furthermore, as pointed out earlier, there were subgroups of responders in the complete diet studies (e.g., Conners et al., 1976; Harley et al., 1978) and other studies (e.g., Goyette et al., 1978) who may be "overwhelmed" in aggregate analyses. Therefore, conclusions regarding the effectiveness of the diet are still not possible. 58 In addition to the weaknesses associated with individual studies, other more general issues that are possibly relevant to the outcomes of some of the studies were also mentioned briefly at different points. These are discussed in more detail below. 1. The dosage of food coloring used in most studies, 1 to 26 mg/day, may be too low. According to Rimland (1983), the FDA found daily consumption of colorings to be 59 mg/day for 1 to 5-year-olds and 76 mg/day for 6 to 12­ year-olds. These are average figures and it is possible that for certain groups of children, particularly among younger ones, consumption may be a lot higher. The 90th percentile consumption figures were 121 mg/day for ages 1 to 5 years and 146 mg/day for ages 6 to 12 years. Maximum consumption was estimated to be 312 mg/day. Rowe (1988) used 50 mg/day and elicited substantial behavioral changes. Likewise, Swanson and Kinsbourne (1980) used 100 and 150 mg/day doses and also obtained significant effects. In addition to this question of inadequate doses, the intervals at which the dyes were administered in the challenge studies may not approximate the naturally occurring frequencies of exposure. Many children consume foods with artificial coloring throughout the day. More continuous exposures may produce more evident effects. 2. Chocolate bars and cookies may be unsuitable as challenge vehicles and placebo items. Chocolate has long 59 been recognized as one of the more common offenders and a variety of adverse reactions to chocolate have been reported (e.g., Egger, Carter, Graham, Gumley & soothill, 1985; Hall, 1976; Hannington, 1980; Randolph & Moss, 1980; Tryphonas & Trites, 1979). SYmptoms included urticaria, fatigue, migraines, mental confusion, and depression. 3. Conners Questionnaires were used in most of the studies to measure effects. Beside being sUbjective, it may not be sensitive enough or the most appropriate instrument. It was noted earlier how in the Mattes and Gittelman-Klein (1978) study, the mother who was able to correctly identify challenge phases beyond chance levels but the Questionnaire ratings did not reflect any effect. Irritability, which was to her the primary reaction, was not measured by the Questionnaire. Also, in some studies (e.g., Goyette et al., 1978), effects were not detected in parents' and teachers' ratings but were evident in the more sensitive measures such as tests of attention. 4. Children who have been on the Feingold K-P diet may have recovered or developed a resilience and be less prone to react to challenges than if they had not been on the diet. It was noted earlier how in many of the studies (e.g., Harley, Matthews, & Eichman, 1978; Rose, 1978; Weiss et al., 1980), children were chosen for participation because they were on and had reported favorable responses to the diet. The children selected may have, therefore, 60 been healthier because the Feingold K-P diet eliminates many junk foods and the children may have been eatin9 more nutritious foods and/or because they experienced a reprieve from substances to which they had cyclic "allergies" (Rinkel, Randolph, & Zeller, 1951; Rowe & Rowe, 1972) and/or because the diets provided relief from constant exposures to particular offenders and enabled them to again tolerate these "allergens". 5. Symptoms may only occur as a result of additive effects. Some (e.g., Tryphonas & Trites, 1979) have suggested that exposure to a single offending food may not result in the expression of symptoms but simultaneous exposure to different offenders may. An individual could possess weak sensitivities therefore making it safe to consume the offenders at different times. Problems may arise only when these foods are eaten together or within a short period of time. Reactions may have occurred in some of the food additive studies if challenges with combinations of offenders were conducted. 6. For the most part, the possibility of synergistic effects were also not considered. BoIsen (1985) noted that it may be that for some, reactions are a result of interactions between offending foods and/or agents. For example, symptoms may not occur with separate exposures to tartrazine and BHT but do occur to foods containing both. The complete investigation of the effects of foods and food 61 substances requires that this possibility also be addressed in challenge study designs. In summary, experimental studies on the behavioral effects of foods and particular chemicals eliminated in the Feingold K-P diet have yielded mixed and inconclusive results. Some of the reasons for the varied results were discussed above. It must also be borne in mind that the foods and chemicals studied are only a fraction of the average diet and a thorough investigation of the effects of food must go much further. These studies have, however, provided valuable insights regarding study designs in research investigating the impact of food on behavior. Clinical Ecology A group that has been very involved in investigating, treating, and reporting the effects of food on behavior are the clinical ecologists. This group was initially composed primarily of physicians who did. not restrict themselves to the narrow, immunologic definition of allergy. They were, therefore, considered a "fringe element of the medical community" (Taylor, 1986, p. 599; staudenmayer & Selner, 1987) who were divorced from the field of orthodox allergy and mainstream medicine. They ascribe to the original, broader definition of allergy as "hypersensitivity to a specific substance (food, dust, pollen, etc.) which in similar amounts is harmless to most people (Crook, 1975). Therefore, earlier in this century, as traditional 62 allergists began excluding food and other environmental exposures as valid areas of study, clinical ecologists continued to actively investigate and treat patients for food and chemical reactions. They were and remain interested in all types of adverse reactions to food. Ecologically-oriented physicians such as Coca (1942), Forman (1932), Hansel (1933), Lee (1961), Rea and Ross (1989), Rinkel, Randolph, and Zeller (1951), and Rowe (1928, 1931, 1950) published many books and articles on the effects of food on behavior. They have linked an astounding variety of physical and mental symptoms to exposures to common foods. A list of foods that they (Dickey, 1976a; Golos & Golbitz, 1979; Kenyon, 1986; Mandell & Scanlon, 1979; Rapp, 1979a, b; Rowe & Rowe, 1972) claim have been demonstrated to elicit reactions in those susceptible are presented in Table 2. Symptoms that have supposedly been reproduced in patients via testing and a comprehensive listing of what they report to be typical symptoms appear in Tables 3 and 4, respectively. They say that reactions to offending foods range from the very subtle, almost undetectable, to the very obvious, even dramatic. They believe that every tissue and organ in the body can be affected. The writings and practices of this group have also stirred much controversy and pUblic interest. Unfortunately, like the early years of the Feingold K-P Table 2. Offending Foods 63

Animal Products Fruit Vegetables Other beef apple asparagus almonds butter apricot beets artificial cheese avocado broccoli sweeteners chicken banana cabbage bakers yeast crab cherry carrot beer egg dates cauliflower black pepper fish grapes celery cane sugar ice cream grapefruit chives chocolate lamb lemon corn coconut lobster lime green beans coffee milk melons lettuce garlic pork orange mushrooms hard liquor shrimp peach onion honey turkey pear potato peanuts pineapple spinach soda pop plum squash soy products Grain sweet tea potato vinegar oats tomato water rice wine rye wheat 64

Table 3. Sympt~ms Reprodu~ed in Patients by Testing for Allergies .. ~ . 1. SKIN: Itching, burning, flushing, hot flashes, warmth, coldness, tingling, sweating behind neck, hives, blisters, blotches, red spots, pimples. 2. EAR, NOSE, THROAT: Nasal obstruction, 'sneezing, nasal itching, runny nose, postnasal drip, sore or dry or tickling throat, clearing throat, itching palate, hoarseness, hacking cough, fullness or ringing or popping of ears, itching deep within ears, earache with red or normal eardrums, intermittent deafness, loss of some tones, sounds much louder, fluid accumulation in the middle ear, dizziness, vertigo, imbalance. 3. EYE: Blurring of vision, temporary loss of vision, double vision, spots before the eyes, pain in or.behind the eyes, watery eyes, excessive tear secretion, crossing 'of eyes, glare hurts eyes, colors look brighter; eyelids twitching, itching, drooping, or swollen; redness and swelling of the lids. 4. RESPIRATORY: Shortness of breath, tightness of chest, not enough air getting into the lungs, wheezing cough, mucus formation in bronchial tubes, rattling sounds or vibrations in the chest. 5. CARDIOVASCULAR: Pounding heart, increased heart rate, skipped beats, flushing, hot flashes, pallor; warmth, coldness, tingling, redness or blueness of hands; faintness; pain in front of the heart; pain in the left arm, shoulder, neck, and jaw traveling down to the wrist (pseudo-heart attack pain). 6. GASTROINTESTINAL: Dryness of mouth, hunger, thirst, increased salivation, canker sores, metaallic taste in mouth, stinging tongue, toothache, burping, retasting foods, ulcer symptoms, heartburn, indigestion, infantile colic, nausea, vomiting, difficulty in swallowing, rumbling in abdomen, constipation, abdominal pain, spastic colitis, "emotional" colitis, gall bladder colic, cramps, diarrhea, passing gas, mucus or blood through the rectum, itching or burning of rectum or anus. 7. GENITOURINARY: Frequent, urgent, or painful urination; inability to control bladder; bedwetting; vaginal discharge; itching; swelling, ~edness or pain in the genitals; painful intercourse. B. MUSCULOSKELETAL: Fatigue, generalized muscular weakness, muscle pain, joint pain, joint swelling with local redness, stiffness, joint deformity, arthritis, soreness, chest pain, backache, neck muscle spasm, shoulder muscle spasm, generalized spacticity, limping gait, limitation of motion. 9. NERVOUS SYSTEM: Headache, migraine, compulsively sleepy, drowsy, groggy, confused, dizzy, loss of ~alance, staggering gait, slow, Sluggish, dull, unable to concentrate, depressed, crying; tense, angry, irritable, anxious, panic, stimulated, aggressive, overactive, frightened, restless, manic, hyperactive, with learning disability, jittery, convulsions, head feels full or enlarged, floating sensation; silliness, poor memory, variations in penmanship legibility, feeling of separateness or apartness from others, amnesia for words or numbers or names, hallucinations, delusions, paranoid ·state, stammering, or stuttering speech, claustrophobia, paralysis, catatonic state, perceptual dysfunctions, typical symptoms of mental retardation. (Mandell & Scanlon, 1976, pp. 16-17) "

65

Table 4. Typical Adverse Reactions to FOods A. Headache F. Gastrointestinal I. Cerebral depression 1. Migraine 1. Cheilitis (inflammation 1. Acute and chronic 2. Vascular of lips, sores at depression 3. Histamine corners of mouth) 2. Drowsiness 4. Tension 2. Aphthous stomatitis .approaching 5. "Emotional" (cold sores) narcolepsy 6. Muscle spasm 3. Aerophagia (air 3. Episodic dullness B. Ophthalmic swaLlowing and belching) or dreaminess 1. Eye pain, itching 4. Nausea 4. Learning disorders 2. Photophobia (pain 5. Vomiting 5. Tension-fatigue from light) 6. Heartburn syndrome 3. Episodic blurring 7. Indigestion 6. Minimal brain 4. Transitory 8. Gassiness dysfunction refractive changes 9. Abdominal pain J. Cerebral stimulation 5. Tearing 10. Cramps 1. Restlessness 6. Allergic shiner 11. Diarrhea 2. Nervousness, 7. Puffy lids 12. Pruritus ani (anal anxious, inner 8. Red, congested itching) shakey feeling blood vessels 13. Irritable bowel 3. Jitteriness, C. Otologic 14. Spastic colon tremor 1. Serous otitis 15. Mucosa colitis 4. Insomnia (fluid in ears) 16. Nervous stomach 5. Hyperactivity 2. Tinnitus (ringing) 17. Food intolerances 6. Behavior problems 3. Meniere's syndrome 18. Bloating 7. Inappropriate 4. Hearing loss 19. Belching, passing gas emotional'outbursts 5. Vertigo 20. Repeating a taste 8. Uncontrolled anger 6. Excessive ear wax 21. Excessive salivation 9. Fear, panic D. Respiratory G. Dermatologic K. Psychiatric 1. Laryngeal edema 1. Urticaria (hives or 1. Feelings of 2. Asthma welts apartness, "spacey" 3. Postnasal discharge 2. Atopic dermatitis (unreal) 4. Allergic tracheitis (eczema) 2. Floating sensation 5. Allergic laryngitis 3. Neurodermatitis 3. Episodic amnesia 6. Allergic rhinitis 4. Adult acne -4. Pathologically 7. Night cough 5. Erythema multiforme poor memory 8. Allergic bronchitis 6. Skin lesions of 5. Inability to E. Cardiovascular porphyria concentrate 1. Extrasystoles 7. Hand dermatitis 6. Personality 2. Tachycardia (rapid 8. Nondescript syndromes changes, psychoses, heart) H. Muscular schizophrenia 3. Palpitation 1. Muscle spasm 7. Autism .4. Episodic syncope 2. Muscle pain 8. Hallucinations (fainting) 3. Muscle cramps L. Urological 5. Generalized 4. Muscle weakness 1. Frequency angio-edema 5. Nuchal pain or 2. Dysuria (painful 6. Angio-edema of rigidity (back) urination) lungs 6. Undue fatigue 3. Nocturia 7. Flushing, 7. Sluggishness (urinating at chilling, 8. Arthritis; joint pain, night) hot flashes, swellings, stiffness 4. Enuresis (poor pallor 9. Rheumatism bladder control, 8. Night sweats 10. Backache bed wetting) 9. Skipped heart beats 5. Genital itching 10. Chest pain or pain 6. Urgency M. Hematological 1. Anemia 2. Neutropenia (decreased white blood cells) 3. Purpura (hemorrhagic spots in skin)

(Mandell & Scanlon, 1976, pp. 17-20) 66 diet, the literature is composed primarily of uncontrolled clinical observations and anecdotal reports (David, 1985; Kettelhut & Metcalfe, 1987; May, 1985, 1986; Pearson, 1986; Terr, 1986). In response to critics, clinical ecologists have replied that more double-blind procedures have not been conducted because most ecologists are clinical practitioners who are providing services to patients and not scientists working in well-funded research centers (Randolph & Moss, 1980; Rossiter, 1985). Rapp (1981) and Randolph and Moss (1980) pointed to results of some single and double-blind studies to substantiate their claims. Unfortunately, quite a few of these are reported in popular books and texts such as Mandell and Scanlon (1979), Randolph and Moss (1980) and Dickey (1976a). They are therein usually described very sketchily with inadequate information for evaluation and/or replication. There have also been some videotapes of trials produced by various clinical ecologists and others (e.g., IIFood Allergy War"). These are fascinating but, again, difficult to evaluate. Some of the studies that have appeared in journ~ls will be described next. Miller (1977) conducted a double-blind crossover study to investigate the efficacy of food extract injection therapy. Evidence supporting the effectiveness of this procedure (King, Fadal, Ward, Trevino, Pierce, Stewart, & 67 Boyles, 1988; King, Rubin, Fadal, Ward, Trevino, Pierce, stewart, & Boyles, 1988; Lee, 1961; Rinkel, Lee, Brown, Willoughby, & Williams, 1964) and refuting it (American Academy of Allergy, 1981; Crawford, Lieberman, Harfi, Hale, Nelson, Selner, Wittig, Postman, & Zietz, 1970) have been reported. Miller first identified the food sensitivities of each of 8 subjects (males and females, ages 4 to 57 years), determined neutralizing doses, and combined these doses into a single solution for each. The active and placebo (phenolated saline diluent) solutions were self­ administered sUbcutaneously in alternating 2o-day courses. During the study, the sUbjects ate only foods for which they were receiving neutralizing doses in the treatment solution. The measures were ratings of improvement in intensity, duration, and frequency of sUbjects' symptoms. The results showed that the active solutions were significantly more effective in alleviating symptoms. It was reported that in "many cases the response of lifelong, severe, intractable syndromes was rapid and dramatic, often beginning in 3 to 4 days and symptoms often returning within 3 to 4 days after beginning a course of placebo therapy" (p. 191). The importance of the technique in diagnosis of food sensitivities was also discussed. There were several apparent weaknesses associated with this study. First, adherence to the diet was determined by sUbjects' diet diaries. Second, neutralizing doses were 68 combined with a single solution therefore not permitting evaluation of specific offenders. Third, the outcome measures were very sUbjective. Rapp (1978b) studied a 9-year-old boy who appeared to be sensitive to many foods. While on a milk-free diet, double-blind milk challenges were conducted with powdered milk and a corn starch placebo administered in capsules. She reported that the child was able to correctly identify when he had taken the milk capsules because of the ensuing sYmptoms which included irritability and the placebo capsules because he remained symptom-free. In another phase, a double-blind test of sublingual milk therapy was conducted. Buffered saline was used as the placebo. The child was allowed to drink milk throughout this phase. It was reported that the mother was able to identify the placebo conditions within 2 days because her son experienced somatic symptoms and became irritable and emotional. She also reported that with the active doses, these symptoms were relieved in 2 days and disappeared in 4 days. Rapp concluded by saying that the techniques she utilized were important in the diagnosis and treatment of food sensitivity but pointed out that they were at the present time no adequat~ explanations of the mechanisms underlying this procedure. Rapp (1979b) published another study in which 10 boys and girls from an earlier study (Rapp, 1978a, described 69 earlier) plus a sibling of one of the sUbjects were treated with food extract. After being on individualized elimination diets, these 6 to 15-year-old children were tested via an intradermal titration method for 10 to 30 common foods to identify sensitivity and to determine treatment (neutralizing) doses. The parents of the children were given three solutions (two placebo and one active) and instructed to administer each either sUblingually or intradermally for 5 to 7 days. Measures were parents' global impressions and Parent Abbott Hyperkinesis Index scores. Improvement on the Abbot scores were obtained for ·all of the children who completed the double-blind phase (i.e., 5 of 8 children). Only one parent's global impressions reflected improvement on the extract treatment. In 5 cases, the mother and children correctly identified the extract solutions and in 2 cases the children correctly identified the treatment but the mothers did not. Lehman (1980) conducted a double-blind study to assess the efficacy of sublingual provocative food testing on 15 SUbjects, ages 1 to 23 years, who had experienced adverse reactions to various foods. SUblingual food drops of egg, corn, milk, yeast and a placebo of distilled water were administered and repeated a month later. To measure the responses more objectively the author relied predominantly 70 on changes in the degree of swelling and edema in the nasal mucosa. The results showed that positive reactions occurred just as frequently to the distilled water as to the food extracts. Lehman concluded that more well-controlled, double-blind studies needed to be conducted. O'Shea and Porter (1981) tested 15 hyperactive children (ages 5 to 13 years) in a double-blind format in an attempt to determine whether certain foods, food colors, and inhalant allergies can cause hyperactive symptoms, and if sUblingual therapy can be an effective treatment. The children were first tested via intradermal and sUblingual provocation techniques to determine sensitivity and neutralizing doses. Foods tested included milk, corn, egg, Wheat, chocolate, inhalants included dust, mold, and tree pollen; dyes included were red, yellow, and blue. As in previous studies, children were allowed to eat the foods being tested. Each child received either the active extract or placebo. Parents kept logs of behavioral and physical symptoms and both parents and teachers were interviewed once a week by a psychologist. Reactions were obtained and neutralized for all allergens tested with the highest responses being to all three dyes (80%), milk (73%), and peanuts and tomatoes (47%). Nine of the 15 children's behaviors were reported improved during the 3 weeks that they were receiving the 71 active extract. Three children were rated as worse during the treatment and three others showed no change during the study. The authors offered possible explanations for each of these six cases. They concluded that the study provides evidence that hyperactivity can be caused by various foods, food colors, and inhalant allergens and that reactions can be treated by extract therapy. However, more objective and reliable measures should have been used in the above studies, for most of the measures were subjective and questionable. King (1981) administered food extracts sublingually to 30 patients at a clinical ecology clinic in a double-blind placebo format. Each of these patients, ages 17 to 56 years, complained of at least one psychological sYmptom (e.g., anxiety, depression). Dependent measures included self-report of symptoms, pUlse rate, signature size, Bender-Gestalt test, Mood Affect Adjective Check List, WAIS, and the Digit Symbol Substitution Test. A repeated measures design with four conditions was utilized: allergen trials, placebo trials, baserate trials (no sublingual administration), and screening trials to assess reactivity to the placebo. The foods tested included Wheat, beef, milk, and sugar. Reported rates of cognitive-emotional sYmptoms were significantly higher for allergens than placebos with the placebo sYmptoms being equal to baserates. There were no 72 other significant differences on the other measures. The author indicated that the data provided some evidence of a link between psychopathology and "sensitivities" to foods. Concerns can be raised, however, regarding the lack of any corresponding significant results on the other measures and the fact that the sUbjects weren't selected because of evidence of reactions to foods. Finally, although not designed blind, O'Banion and Greenberg (1982) conducted an interesting study that is worthy of mention. A 33-year old female who presented a history of hyperactivity, migraine, perceptual and learning problems, and depression was exposed to diet phases of sensitive and nonsensitive foods (i.e., an A-B-B-A-B-A design). In the A phases she ate only safe foods. In the B phases she ate foods that had been determined to be offenders via earlier provocative testing. Each phase lasted 6 days. The consecutive "B" phases were meant to prevent cyclic effects and permit evaluation of cumulative effects. Measures included responses to affect checklists and performance on coding and hand-eye coordination tasks. Data were obtained at 6:00 a.m. and 6:00 p.m. daily. Visual inspection of the graphed results showed that during the B phases, performance on the academic and motor tasks deteriorated and that depression scores increased. In addition, as would be predicted by clinical ecologists, during the B phases, the sUbject reportedly experienced 73 stronger urges to eat and gained weight. The authors concluded that the results indicated "a strong relationship between both behavioral and physiological changes and the different dietary phases". The major weakness of this study was, of course, the non-blind design. The sUbject was very aware of the different phases. Measures by a blind observer could have easily been incorporated into the procedures and would have provided more objective data. In summary, these studies provide some evidence that foods do impact behavior. Again, however, because of the inconsistent results and methodological limitations of these studies no definitive conclusions can be drawn. Food Allergy As discussed above, historically much of the interest in the adverse reactions to foods has existed outside the mainstream of modern medicine. Recently, however, there has been expanding interest in the traditional medical community. There are increasing number of articles in orthodox-oriented journals dealing with the immunologic and nonimmunologic effects of foods. Unfortunately, many of these studies have not been conducted under controlled double-blind conditions (e.g., Jacob & Carron, 1987; Gettis, 1987; Wall, 1987). These have also been primarily focused on classic and other somatic symptoms. Only a few 74 have been concerned with behavioral reactions. The following is a review of the better designed studies. Maslansky and Wein (1971), in a study of chocolate intolerance, first surveyed 500 allergy patients and found that 81 felt that they were allergic to chocolate. From this group of 81, they eliminated those who couldn't swallow capsules, those whose symptoms were too vague or inconsistent, and those who experienced constant allergic symptoms and were left with 10 SUbjects (ages 4 to 60 years). The allergic symptoms reported by SUbjects included severe headaches, sneezing, tension-fatigue syndrome, hives, Wheezing, rhinnorhea, nausea, and cramps" (p. 8). In a double-blind challenge with capSUles of pure de-fatted cocoa and an unspecified placebo, these SUbjects took six active or placebo capsules each morning after breakfast for 6 consecutive days. According to the authors, six capsules of cocoa contained more cocoa than two average 3/4 ounce bars of packaged milk chocolate. The allergic symptoms were reproduced in 3 of the 8 cases who completed all phases of the study. These SUbjects experienced fatigue syndrome, hives, nausea, and cramps. Unfortunately, as in many other studies, evidence of reactions were obtained from SUbjects' food and symptoms diaries and post-treatment interviews. Check (1983) in JAMA, reported on a study conducted in Denver by S. Allan Bock in which 98 children with possible 75 food-related problems were examined. Symptoms reported included stomach upset, skin rash, respiratory problems, and irritability. Only 11 of the 98 possible cases were verified by double-blind food challenges. Another 29 were considered to be "probable" cases, for although not confirmed in a double-blind trial, there was strong evidence of it (i.e., "reaction reported in an unequivocal way on more than one occasion"). Although there were many suspect foods (e.g., Wheat, egg, corn, milk, tomatoes, etc.), reactions were confirmed for only milk, soy, and peanuts. In another study, Bock, Lee, Remegio, and May (1978) studied 68 children between the ages of 5 months to 15 years who had histories of adverse reactions to 14 foods being studied (these included milk, egg, Wheat, cocoa, etc.). The food products were presented in opaque capsules or mixed into another food for those who had difficulty swallowing capsules. Amounts administered ranged from 20 mg to 8000 mg. In the double-blind challenge, reactions occurred in 16 of the 43 sUbjects, 3 years of age or older, and in 13 of 25 children who were younger than 3 years. The most common offenders were milk, egg, peanuts or other nuts, and soy. Reactions listed included eczema, urticaria, vomiting, Wheezing, diarrhea, and abdominal pain. No behavioral responses were reported. However, in a more 76 recent article (Bock, 1986), one of the authors stated that behavioral abberations had been noted along with the somatic complaints. He did say, too, however, that the behavioral manifestations were never the sole symptoms of the adverse reactions. Crayton, stone, and stein (1981) reported an extraordinary case in which a 19-year-old female who suffered epileptic seizures was tested in double-blind placebo controlled trials. A preliminary neurological examination and hematological workup were within normal ranges. She did, however, react positively to skin tests for various foods and airborne excitants. Via food testing with an elimination diet they discovered that she reacted with tachycardia, hypothermia, and seizures after eating beef and several other foods. To verify these reactions she was given 6 grams of chopped beef or chicken (she had not reacted to this during the elimination diet) in a gelatin capsule. The capsules were administered in a random, double-blind sequence. Pulse, peripheral skin temperature, and EEG recordings were monitored. The chicken did not produce significant changes in any of these measures. Ho~ever, 10 minutes after the beef capsules were given, her pUlse increased from 70 to 85, she became unresponsive to physical and verbal stimuli, and at 50 minutes she experienced the first of 6 grand mal seizures. "The EEG showed only motor activity which 77 completely obscured the recording" (p. 194). Two hours later she was given chicken again and no significant changes in the measures were noted. About an hour after that she was challenged again with the beef. In 15 minutes her pulse increased from 74-80 to 120-124, skin temperature dropped from 34 C to 26 C, and the first of 8 more grand mal seizures began. EEG recordings again only showed motor artifact. The authors reported that sUbsequent to these trials, while on a diet that eliminated her food offenders and anticonvulsant medication, she had experienced only one seizure in 6 months. In the 6 months prior to study, she had experienced 31 seizures despite being on anticonvulsant medication. In a more recent article, Check (1983) reported that the woman had maintained the elimination diet and not suffered a seizure in almost 2 years. Farah, Calder, Benson, and MacKenzie (1985) studied gastrointestinal symptoms in a sample of 49 men and women who showed evidence of food intolerance. They were first put on a low allergenicity diet. The responders were then tested weekly by the reintroduction of single foods to identify the specific offenders. Offending foods were identified in 8 sUbjects who were then challenged with these foods in a double-blind placebo format. Suspected foods were confirmed as causes of symptoms in three 78 individuals. One of these sUbjects reported irritability as the primary complaint. It is interesting to note that the other controlled double-blind studies, which were primarily focused on physical symptoms, strongly indicted food as an etiological factor (e.g., Aas, 1978; Atherton, 1985; Atherton, Sewell, Soothill, Wells, & ChiIvers, 1978; Atkins, steinberg, & Metcalfe, 1985a, b; Bernstein, Day, Welsh, 1982; Bock, 1982; Dahl, Henriksen, & Harving, 1986; Egger, Carter, Wilson, Turner, & Soothill, 1983; Ford & Fergusson, 1980; Jakobsson & Lindberg, 1983; Leinhas, McCaskill, & Sampson, 1987; Mansfield & Bowers, 1983; Mansfield, Vaughan, Waller, Haverly, & Ting, 1985; May, 1976; Onorato, Merland, Terral, Michel, & Bousquet, 1986; Pelikan & Pelikan-Filipek, 1987; Sampson, 1983; Sampson & Albergo, 1984). Ten of these articles were published in the Journal of Allergy and Clinical Immunology which is the official pUblication of the American Academy of Allergy and Immunology whose membership is predominantly orthodox allergists. These articles are evidence that although food is now being studied more earnestly as a source of adverse reactions, they are still primarily interested in the classic sYmptoms and in investigating and verifying immunologic mechanisms. According to May (1984), based on scientific studies, the only manifestations of true food sensitivity are vomiting, diarrhea, fatty stools, edema, rash, hives, and 79 eczema. However, Dr. John Anderson of the Henry Ford Hospital at the Marabou (Sweden) symposium on Food sensitivity (Nutrition Reviews, 1984) correctly stated that "allergists are not necessarily the best to evaluate behavioral abnormalities and it probably takes a team approach of very qualified people for evaluation••. " (p. 121). Sugar Studies Another dietary factor that has received some attention is sugar consumption. Recently, there has been much written in the popular literature about the supposed adverse effects of sugar on the behavior of children, especially those who are hyperactive/overactive. Growing numbers of parents and teachers believe that sugar negatively impacts the social and academic behavior of hyperactive children (McLoughlin & NaIl, 1988; Milich & Pelham, 1986; Wolraich, Milich, Stumbo, & Schultz, 1985). Furthermore, a survey of pediatricians and family practitioners in Washington conducted by Bennett and Sherman (1983) revealed that 45% of pediatricians, family practitioners, and general practitioners recommended at some time a low sugar diet when treating these children. Results on the studies of the Feingold K-P may offer some support for this position because since the diet eliminates all foods with artificial coloring and flavors and there­ fore, probably decreases sugar consumption, some of the 80 reported improvements in hyperactive symptoms may possibly be sugar-related. However, as has been pointed out by many (e.g., Ferguson, Stoddart, & simeon, 1986; Wolraich, stumbo, Milich, Chenard, Schultz, 1986; Varley, 1984), there is very little empirical evidence of such a relationship. Prinz, Roberts, and Hantman (1980) conducted a correlational study focused on the relationship between hyperactive and control children's diets and behavior. Seven-day dietary records including amounts of sugar consumed were obtained for 28 hyperactive and 26 control children ranging in ages from 3 years, 9 months to 7 years, 11 months. The intake levels were compared to children's playroom ratings of destructive-aggressive behavior, restlessness, and quadrant changes which were recorded by "blind" observers through a one-way mirror. The results showed no significant differences between the two groups in amounts of sugar consumed. There were, however, significant positive correlations between the amount of sugar consumed and both destructive-aggressive behavior and restlessness for the hyperactive children. There was also a significant positive correlation between sugar intake and number of quadrant changes for the control children. Wolraich, Stumbo, Milich, Chenard, and Schultz (1986), cautioned that the method of calculating sugar consumption 81 used in the above study was based on weight of the food rather than the weight of the nutrients and this may have resulted in erroneous estimated intake levels. For example, this method could result in a higher score for foods with high water content (e.g., a sweetened beverage) than a highly sugared, but light pastry. They, therefore, conducted a replication study using the more accurate method of calculating sugar consumption. Dietary records including the intake levels of sugar, carbohydrates, proteins, and fats were obtained. Behavioral observations were gathered on 37 variables including assessments of activity, impulsivity, and compliance during play and classroom tasks, laboratory measures of impulsivity and attention, and learning tasks. In addition, they evaluated parents' success at restricting their children's sugar intake. As in Prinz, Roberts, and Hantman (1980), the 3-day dietary records of their 32 hyperactive children (ages 7 to 12 years) were not significantly different from those of 26 matched control children. Also similar to the earlier study, significant positive correlations were obtained for the hyperactive children between three free-play behavioral measures (i.e., ankle actometer, grid crossing, attention shifts) and sugar and carbohydrate intake. A significant negative correlation was found for on-task behavior. They also reported that more parents of hyperactive boys 82 indicated attempts at restricting their children's sugar intake. The authors concluded, however, that parents who were attempting to restrict sugar intake were not necessarily more successful than those who were not. The authors stressed caution in interpreting these results because: (1) only 4 of the 37 behavioral measures were significant and these differed from those reported in Prinz et ale (1980), (2) most intervention studies (to be described next) have not found significant effects, suggesting that the greater the level of hyperactivity, the more sugar consumed and not vice versa, and (3) the method of obtaining dietary information may not have been reliable (i.e., there was low correspondence between the hyperactive child and mother diet reports). In addition to these concerns, observational data were not obtained on the control group because of funding restrictions. Obviously, causality cannot be determined by correlational studies like the two above. Fortunately, some experimental studies have been conducted. Two double­ bl~nd crossover studies by Conners, Wells, Horn, Blouin, Beerbohm, O'Donnell, Seidel, and Shaw (1982) were described by Ferguson, Stoddart, and Simeon (1986) and Wolraich, Milich, Stumbo, and Schultz (1985). In both studies, inpatients of a child psychiatric ward were challenged with sucrose, fructose, and aspartame (Nutrasweet). In the first study (N=12), no attempt was made to control the 83 children's sugar intake and the challenge drink was served in the morning at breakfast. The measures were activity level, sustained attention levels, and degree of compliance on the ward. There were significant increases in the children's total motor activity when challenged with either 50 gm of fructose or sucrose as compared to aspartame. In the second study (N=37), 1.25 gm/kg doses of sucrose significantly reduced fine motor activity and similar doses of fructose significantly reduced gross motor activity. Similarly, Behar, Rapoport, Adams, Berg, and Cornblath (1984) in a double-blind study, tested twenty-one 6 1/2 to 14-year-old boys with glucose and sucrose. These children were considered by their parents to be adversely affected by sugar and constituted a heterogeneous group (i.e., 9 were hyperactive, 4 had previously suffered from obsessive-compulsive disorder, 8 did not qualify for a psychiatric diagnosis, etc.). The children received on 3 days (a minimum of 48 hours apart) either saccharin, saccharin and glucose (1.75 gm/kg) or saccharin and sucrose (1.75 gm/kg). Measures included Conners Questionnaire scores, readings from an acceleration measuring device, scores on memory and attention tests, and blood analyses. The results showed only a slight but significant decrease in observed motor activity with the sucrose challenge. The glucose and saccharin drinks did not result 84 in any significant changes in any of the behavioral, cognitive or physiological measures. These results are difficult to interpret because of the heterogeneity of the sample and the possibility of interactive and/or overwhelming effects of the saccharin. Wolraich, Milich, stumbo, and Schultz (1985) conducted two studies in which hyperactive boys were admitted to a clinic and challenged in double-blind fashion on 2 consecutive days with either sucrose (1.75 gm/kg) or aspartame. Measures included playroom observations and a variety of cognitive tests (e.g., paired-associate learning test, continuous performance test, etc.). All sixteen 7 to 12-year-old boys showed no significant differences between the sucrose challenge and aspartame on any of the dependent variables. The authors noted that since the challenge was conducted an hour after lunch, any effects of the sucrose may have been attenuated. Therefore, in the second study, 16 other hyperactive boys (at least 8-years-old) received the challenge drink by itself in the morning after an overnight fast. As in the first study, there were no significant differences between the sucrose challenge and aspartame. Milich and Pelham (1986) used an almost identical procedure. Their sample of 16 hyperactive boys (ages 6.4 to 9.1 years) fasted overnight and were challenged in the 85 morning with a 1.75 gm/kg sucrose drink or aspartame. The study was conducted over 4 days with the boys randomly receiving sucrose or aspartame on 2 days each. Dependent variables included measures of playroom behavior, academic accuracy and productivity, and noncompliance with adults. Consistent with the other challenge studies, there were no significant differences between the sucrose and aspartame phases. The author discussed several possible explanations for the lack of effect: (1) the effects of the sucrose may have been masked by the rigorous behavioral treatment program that the hyperactive sUbjects were experiencing concurrently, (2) small sample size, and (3) doses may have been to small relative to normal daily intake (this also applies to the three previous studies). Ferguson, Stoddart, and Simeon (1986) conducted two studies with the first designed to examine the behavioral and cognitive effects of a range of doses of sucrose. Eight children, ages 5 to 13 years, who were believed by parents to react adversely to sugar, were challenged in a double-blind format with three administrations of sucrose and three of aspartame. Each administration was separated by at least 48 hours to avoid any possible carry-over effects. Three dosage levels (i.e., .5 gm/kg, 1.0 gm/kg, and 1.5 gm/kg) were randomly assigned as well as sucrose and aspartame administrations. Single sUbject analyses and 86 t-test comparisons provided no evidence of any differences between the sucrose or aspartame phases. The second study was conducted to test younger children than previous studies (ages 3.5 to 5.3 years). The children were again challenged in a double-blind format with either sucrose or aspartame and behavioral and cognitive measures were obtained. Results showed that 3 of the children scored consistently lower on drawing tasks on sucrose days and 3 others showed marked decreases in measured activity on sucrose days. The authors pointed out that the deterioration on the drawing task was not paralleled by similar changes on any other measure (not even on another coordination task) and that the decrease in activity was not what most would expect. In one of the better studies in this area, Goldman, Lerman, Contois, and Udall (1986) tested 8 normal, healthy preschool children with sucrose and aspartame. The children again fasted overnight and drank juice sweetened with 2 gm/kg of sucrose (similar to the amount of sugar found in a 12-oz. can of soda) on one morning and juice "sweetened to an equivalent degree" with aspartame (250 mg/kg) on another morning a week later. The order of the drinks were random and all directly involved with the children were blind in regard to the type of dirnk. After the challenges, the children were observed in structured- 87 task and free play situations. In the structured sessions, measures were obtained on ability to sustain attention and inhibit both gross motor and fine motor movement. In the free play sessions, the children were scored as exhibiting either appropriate or inappropriate school behavior based on whether they were stationary or moving and focused on a task or distracted. The results showed that on the sustained attention measure where the testing occurred before the drink and 30 and 60 minutes after the drink, the children's errors increased over time in the sucrose condition and decreased in the aspartame condition. In addition to this significant drink by time interaction, there was also a significant difference between conditions in regard to the number of errors committed during the 60 minute testing sessions. No significant results were obtained for the other measures. In the free play sessions, there were significantly more movement and time off-task following the sucrose drink. The greatest differences in these two meausres between conditions occurred 40 to 55 minutes after the challenges. Given these results, the children's behavior during these sessions were· obviously also found to be significantly more inappropriate following the active challenges. Again, the greatest differences in 88 inappropriate behavior between conditions occurred 40 to 55 minutes after the challenges. The authors concluded with a discussion of how the method and results of this study may clarify why so many other experimental studies have found no significant responses to sucrose challenges. They suggested that for several of the studies reviewed earler, this may have been due to one or more of the following: a) the procedure for data collection and analyses that they used in which data was aggregated over 15-minute intervals may be more appropriate and sensitive because the effects of sucrose apppear to be short-term, b) pre-school children may be more sensitive to sucrose than the school-aged, c) fasting before the trials is necessary for controlling dietary and metabolic status of the sUbjects, d) since the effects of sugar on activity level appear to be less dramatic, these effects may be overwhelmed in children with high baseline activity levels such as with some hyperactives, and e) research settings which are inhibitive and/or unfamiliar may be a confounding factor. The authors also pointed out several issues that need to be considered in interpreting the results of this study. First, it is possible that the significant responses are not specific to sucrose and may also be obtained with other forms of carbohydrates. Second, although there is evidence against it, the effects of the challenge may have been 89 exaggerated because the sucrose drink was taken without other foods. Third, again though there is no evidence of it, aspartame may have improved behavior during the control condition. In a more recent study, Rosen, Bender, Sorrell, Booth, McGrath, and Drabman (1988) examined the effects of sucrose on both preschool and elementary school children. None of the children demonstrated mental deficiency but 7 preschoolers and 1 elementary child scored above 15 on the Conners Teaching Rating Scale which is the cutoff commonly used to identify attention deficit disorder. All children experienced 3 double-blind conditions: high sugar, low sugar, and aspartame (control). The challenges occurred over 15 consecutive school days with 5 days spent in each condition. The condition each day was randomly determined so the 5 days of each condition were not necessarily consecutive days. On each day, the children fasted overnight and received either a drink containing 50 gm of sucrose (high sugar), 6.25 gm of sucrose (low sugar) or no sucrose (control-aspartame) with their breakfasts. Data on cognitive effects were obtained with paired associate learning, matching, and simple academic tasks. Each day, teachers also completed the Abbreviated Conners Teacher Rating Scale and a 10-point global rating scale which was a Likert-type scale intended to measure general changes in behavior such as pace of activiity and 90 disruptiveness. In addition, the preschoolers were observed in free play sessions and the elementary children in in-seat work sessions. Data on fidgeting, activity changes, active movement, vocalizations, and aggression were collected for the preschoolers. Fidgeting and on-task behaviors were recorded for the elementary school children. The results of the cognitive measures showed that the girls made significantly more errors during the high sugar condition as compared to the low sugar condition. The only other significant finding was that the cognitive measures, in toto, suggested differences between preschoolers and elementary school children in their responses to the challenges. Unfortunately, none of the analayses on individual dependent measures were significant making further interpretation of this multivariate effect difficult. The teachers' ratings revealed that scores on the global rating scale were slightly but significantly higher during the high sugar than low sugar condition. Not surprisinging, the observational measures showed that a significantly higher percentage of fidgeting occurred in the in-seat and structured setting of the elementary school than in the free play setting of the preschool. Comparisons between children grouped according to whether they were scored as high or low on the abbreviated Conners Teacher Rating Scale showed no significant 91 differences between performance or behavior in the high and low sugar conditions. The authors felt that although significant results were obtained, the magnitude of the effects was small and the outcomes need to be interpreted cautiously. They concluded that the results were consistent with most of the other studies in that no clinically significant behavioral effects were elicited. The design weaknesses as discussed in the article included: a) doses of sucrose were not titrated to the subjects' weights so smaller sUbjects received proportionately larger amounts of sugar, b) dose schedule was such that it was not possible to pinpoint the specific amount of sucrose needed to produce an effect, c) small sample size, and d) brevity of obsevation periods (2 to 4 minutes each day). One of the obvious strengths of this study was that the challenges were part of a complete breakfast each morning. As pointed out by others (e.g., Goldman et al., 1986), this comes much closer to what would occur in a non-experimental setting. In summary, the results of these challenge studies provide, at best, mixed evidence for an effect of sugar on behavior. However, even those who doubt that any future challenge studies will yield any different results concede that there may be subgroups that are adversely affected {Ferguson, Stoddart, & simeon, 1986; Wolraich, Milich, 92 stumbo, & Schultz, 1985). This possibility of select groups of children reacting (i.e., there were significant positive correlations between sugar intake and activity in the control group in Prinz et al., 1980 and the preschoolers in Goldman, et al., 1986) and that of higher, continuous doses being required to produce more significant effects should be pursued. Diet and Antisocial Behavior Another body of literature that is focused on a facet of the relationship between food and behavior is that dealing with the effects of diet on antisocial behavior. Reed (1977), a probation administrator, claimed that manipulation of diet was an effective means of managing probationers. She, too, was convinced of a strong link between sugar consumption and behavior. Unfortunately, "evidence" for her claims were primarily anecdotal. Since her report, there have been texts (Hippenchen, 1978; Schauss, 1980) and articles (e.g., Rimland & Larsen, 1981) which discuss some nutritional and biochemical mechanisms that may explain the relationship between diet and crime and delinquency. Among the theories are that of reactive hypoglycemia, food allergies/intolerances, food additives, and mineral and vitamin deficiencies. Gray and Gray (1983) and Gray (1986) published critical reviews of these theories and concluded that there was little evidence to support any of these proposals and that for the most part 93 they were based on anecdotes. They did, however, urge further research. Similarly the American Dietetic Association declared in a position statement (1985) that a causal relationship between diet and crime has not been demonstrated. Nevertheless, they concluded that dietary improvements based on firm evidence are desirable. Some of the better studies are discussed next. D'Asaro, Groesbeck, and Nigro (1975) reduced the refined carbohydrates in the diets of 44 prison inmates, instructed them in nutrition, and gave half of them a vitamin supplement pill and the other half a placebo pill. The authors reported that the group that received the vitamin supplement experienced an overall improvement which was reflected in their psychological test scores. As pointed out by Peace and Love (1986), there were several major problems in this study. First, over 50% of the subjects dropped out of the study leaving 16 inmates in the experimental and 7 in the control group. This raised concerns regarding small sample size and differential drop out rates. Second, the psychological test scores of the vitamin and placebo groups who were on the diet and who received nutritional instruction were very similar. Only the inmates that were not on the diet and who did not receive pills or nutrition instruction showed no improvement. Evaluating the effects of the supplements are, therefore, difficult. A stronger argument can be made 94 for the efficacy of the diet. Third, statistical comparisons were made only within groups so there is little basis for between group (i.e., vitamin vs. placebo) comparisons. Finally, 8 weeks may not be long enough to assess the effects of the manipulations. Fishbein (1981) classified 104 adult inmates as either hypoglycemic or nonhypoglycemic based on a symptom checklist and amounts of refined carbohydrates that they usually consumed. Each group was randomly divided into a treatment group that received a low refined carbohydrate diet and a control group that remained on the usual high carbohydrate prison diet. The Hoffer-Osmond Diagnostic Test (HOD; yields paranoid, depression, thought disorder scores, etc.) which were completed by the inmates served as the dependent measure. The results showed a significant improvement only for the hypoglycemics on the low carbohydrate diet. The study, however, was nonblind (i.e., inmates on the experimental diet wore armbands to remind others that they were on a restricted diet). The validity and reliability of the questionnaires used to identify hypoglycemics and the HOD have also not been established (Schoenthaler, 1985). For example, the results of the questionnaires were not compared to results of an instrument such as the Glucose Tolerance Test which measures glucose levels in the blood. 95 Schoenthaler (1982, 1983a, b, c, d, e) conducted a series of studies in 10 different juvenile correctional facilities across the United states. As in some of the studies discussed earlier, he focused primarily on the effects of sugar and refined carbohydrates. At nine of these institutions the diets were modified to reduce the juveniles' intake of sugar and refined carbohydrates. This typically meant replacing "junk food" snacks with more nutritious foods (e.g., vegetables, fruits, popcorn), sUbstituting sugared soft drinks with fruit juices (e.g., orange, grape, grapefruit), and eliminating high sugar-content foods from the diet. Astounding reductions in both serious (e.g., assault) and minor (e.g., refusal-to-obey-orders) incidents were reported. In most of the centers, 40% or greater reductions of antisocial behaviors were noted. In the tenth institution, in addition to a low sugar, low additive diet, orange juice was made more available to an experimental group at meals and for snacks. The juveniles were permitted to consume as much as they desired. The incidence of antisocial behaviors declined 47% from the rate established on the low sugar, low additive diet. These are impressive results, however, as Schoenthaler (unpublished) pointed out, each of these studies have methodological weaknesses. For example, in one study 96 (Schoenthaler, 1982), the before-after design was conducted without random assignment, thereby creating concerns regarding the comparability of the treatment and control groups. In two studies (Schoenthaler, 1982, 1983b), the samples studied were quite small. In the earlier study, there were only 58 subjects (i.e., N=12 for the experimental group and N=34 for the control group). In the more recent study, an important design feature was included, that is, after the experimental condition there was a reversal stage in which the institution reverted to its original diet." Unfortun-ately, there were only 28 sUbjects who acted as their own controls by experiencing both the treatment and baseline or reversal phases. In these two studies, generalization of the results are, therefore, severely restricted. Some of the studies (e.g., Schoenthaler, 1983c, 1983d) did not use double-blind formats. In these studies it is difficult to eliminate the possibility of Hawthorne or placebo effects as explanations for the observed reductions in antisocial behavior. Schoenthaler (1985) recognized, too, that the results of his six studies could also be due to maturation effects. In response to some of the criticism regarding the plausibility of these alternative explanations, he conducted a time series analysis at two Alabama detention 97 facilities. Infraction rates were observed during a one- year baseline phase and a one year diet phase. The number of infractions were at a relatively constant level during baseline, dropped gradually after the dietary intervention, and stabilized after about 20 weeks. He declared that placebo effects would have been reflected in a gradual rise of infractions after the intervention as the effects wore off and that maturation effects would have resulted in improvement during the year-long baseline period. Additional analyses were necessary, however, because the number of cases across conditions were not constant. Some juveniles left the institution at the time that the infraction rates dropped and it may have been that those with the highest rates left and the improvement was a result of their departure. An analysis of the infraction rates of only those who were present during the baseline and treatment still yielded a significant improvement in behavior. In addition to the above, other weaknesses included:

(1). the lack of quantitative information verifying that there was indeed a reduction in sugar consumption as a result of the dietary changes, (2) the possibility that the disciplinary records kept by different correctional personnel were not consistent or reliable, and (3) the failure to use the proper statistical analyses (Gray,

1986) • 98 In all of Schoenthaler's studies, the independent variable (i.e., dietary changes) lacked the specificity necessary to determine why the reduction in incidents occurred. Schoenthaler (1983c, 1985) discussed possible explanations for these effects. The first explanation is that the antisocial behaviors of the juveniles may be due to marginal or subclinical vitamin and mineral deficiencies and the dietary modifications increased the consumption of more nutritious foods, thereby alleviating these deficiencies. The results of the citrus study (Schoenthaler, 1983c) support this hypothesis. Orange juice is rich in essential nutrients such as vitamin C, thiamin, folic acid, and many minerals. Furthermore, ascorbic acid enhances the body's ability to absorb iron and other minerals. The results of the time series stUdy described above provide additional support, for it was shown that improvement in behavior was gradual and appeared to be linear; a pattern that is supposedly indicative of a deficiency etiology. The second explanation is one of endocrinopathy, that is, reactive hypoglycemia. It may be that the regular consumption of high qu~ntities of sugar results in hypoglycemia. In hypoglycemic individuals, the brain may not receive adequate glucose which in turn may result in antisocial behavior. Reducing the amounts of sugar consumed may correct this hyperinsulin condition. The 99 American Dietetic Association (1985) declared, however, that there was no valid evidence to support the position that reactive hypoglycemia causes violent behavior. The results of the citrus study also do not support this hypothesis because greatly increasing the intake of orange juice meant increasing the sugar consumed. If the hypoglycemia explanation were true, antisocial acts shoUld have increased. Schoenthaler (1985) also tested 35 delinquent children in a five-hour glucose tolerance test and fQund that only one of the children was hypoglycemic. Furthermore, he felt that if the improvements were due to alleviation of this condition, infractions should decrease immediately after the diet intervention, for this condition can be brought under control in minutes and stay at that level. The time series study did not show such a pattern. The data from the glucose tolerance test study did, however, show some inverse correlations between blood sugar levels and mood states such as tension and anger, and the ten MMPI scales, thereby offering some support for a low blood sugar hypothesis. It should be noted that there are many cases cited in which those sUffering from food "allergies" are mistakenly diagnosed as hypoglycemic (Mandell, 1981; Mandell & Scanlon, 1979; Randolph & Moss, 1981). The third explanation is that the antisocial behavior may be due to low levels of the neurotransmitter, 100 serotonin, in the brain. Serotonin levels are dependent on the quantity of its precursor, tryptophan. According to Schoenthaler (1985), consumption of certain foods such as orange juice is most effective in elevating tryptophan levels. He, therefore, made a lot of orange juice available to a treatment group and compared their behavior against the behavior of a control group. There were no significant differences and he concluded that this did not support the hypothesis that increased serotonin levels improves behavior. However, weaknesses associated with the design of this study render these results inconclusive. The fourth explanation is that food additives may have been at least partially responsible for the antisocial behaviors exhibited by the juveniles and that the reduction in intake during the changes in diet was responsible for the decreased infraction rates. The foods that were restricted or limited because of high sugar content tend to also contain many additives. Schoenthaler (1985) reported that the results of his time series analyses were inconsistent with this because the resulting pattern should have shown substantial improvement during the second week and a quick leveling off thereafter. The fifth explanation is that the elimination or reduced consumption of different foods, not the sugar or additives per se, were responsible for the improvement in behavior. Among the dietary modifications were replacing 101 sugared cereals with those that were not pre-sweetened, replacing processed foods with fresh produce, eliminating snack foods high in sugar and/or fat, and prohibiting parents from bringing in snacks which were high in sugar content. Plainly, a number of different foods previously consumed were eliminated or their intake reduced. This may have resulted in behavioral changes in those who are susceptible to these foods. In the citrus study, as mentioned earlier, the youngsters drank considerable quantities of orange juice. Although Schoenthaler claimed that milk consumption did not decrease he admitted that it was not carefully monitored. There is the possibility, therefore, that the juveniles drank less milk when orange juice was made more available. Milk is one of the leading food offenders (Randolph & Moss, 1980) and there is evidence linking it to antisocial behavior (Schauss & simonsen, 1979). Most sugar is manufactured from three sources: cane and beets (sucrose) and corn (dextrose and glucose) • Adverse responses to each of these sources have been frequently reported (e.g., Golos & Golbitz, 1979; Mandell & Scanlon, 1979; Randolph & Moss, 1980; Rowe & Rowe, 1972). These reactions range from feelings of anger and apprehension to psychotic-like episodes. Corn is one of the most common offenders in the u.S. (Golos & Golbitz, 1979; Randolph & Moss, 1980). Beets have also been 102 implicated in a number of illnesses (Mandell & Scanlon, 1979). Therefore, adverse reactions to sugar of corn and beet origins appear to be a definite possibility. In regard to cane sugar, the refining process involves exposure to chemical contaminants which are not totally purged from the final crystalline form. This, then, may be another source of reactions. In a double-blind study described earlier, Egger et ale (1985) reported that of 9 hyperactive children who reacted to sugar, 5 reacted to both beet and cane sugar, 3 to cane sugar only, and one to beet sugar only. In addition, according to Golos and Golbitz (1979), regular ingestion of considerable amounts of sugar stresses the neuro-endocrine system and lowers resistance to substances to which individuals are sensitive. With this decreased resistance, symptoms are much more likely to become. manifest. As mentioned earlier, Schoenthaler (1985) stated that the gradual improvement in behavior supported only the marginal malnutrition hypothesis. Similar to food additives, he felt that if antisocial behavior could be explained by a food allergy or intolerance hypothesis, improvement in behavior would have increased sharply and then leveled off during the initial weeks. It must be kept in mind, however, that his data were analyzed in aggregate form and individual patterns were not scrutinized. It still may be that at least some improved because of relief 103 from fixed or cyclic food sensitivities. Furthermore, although withdrawal effects usually arise and subside in 3 to 4 days, Randolph and Moss (1980) reported a case in which it took 6 months to break an addictive-type allergy. withdrawal effects may be experienced until the addiction is broken. In short, an adverse food reaction hypothesis cannot be disputed based on the data reported. Finally, in regard to plausible explanations, it is also possible that all of the above to some degree, a combination of the above, or synergistic effects may account for the reported decreases in infractions. The results of all of these studies provide some evidence of a link between the dietary manipulations and rate of infractions. However, because of weak study designs and inappropriate statistical analyses, the results must be interpreted cautiously. Further research is definitely required (Gray, 1986). In Vitro And Animal Studies The reports of behavioral effects of food coloring and the popularity of the Feingold K-P diet also prompted animal and in vitro studies. These studies provide some evidence to support the claims that food dyes affect behavior. Lafferman and Silbergeld (1979) investigated the effects of erythrosin B (FD&C Red No.3) in different concentrations on synaptosomes prepared from rat caudate 104 nuclei. They found that the dye significantly decreased the dopamine uptake (i.e., absorption back into the nerve terminals from which they were released). Since dopamine (a neurotransmitter) apparently increases activity and decreases sedation (Wender, 1973), this inhibition of uptake may result in higher levels and/or the prolonged presence of dopamine. The authors concluded that this is "consistent with increased dopaminergic activity in vivo, which is suggested to be involved with the hyperkinetic syndrome" (p. 412). Augustine and Levitan (1980) also studied the effects of erythrosine via an in vitro technique. They were interested in the determining whether the dye affected synaptic transmission in frog neuromuscular junctions. Varying concentrations of the dye were applied to the frog's synapses and the results of "conventional intracellular recording techniques" showed increased transmitter release with higher concentrations producing higher rates of release. They concluded that their data indicating pronounced and irreversible effects of the dye on synaptic transmission at low doses support earlier studies suggesting that erythrosine and other food additives can affect behavior. In an earlier study, Levitan (1977) applied xanthane FD&C dyes to invertebrate (mollusk) neurons and found that the physiological characteristics of these neurons were 105 altered. There was a rapid increase in the membrane potential and conductance of the neuron due to an "increase in the conductance of the membrane to potassium relative to the other ions" (pp. 2914-2916). This altered permeability affects neurotransmitter release and, therefore, possibly behavior. The biological activity of artificial food dyes detected in these studies can greatly bolster the argument for behavioral effects. However, as pointed out by some of the authors, generalizing these results to the behavior of animals and humans is not possible until it can be established that these substances do cross the blood-brain barrier (Augustine & Levitan, 1980; Lafferman & Silbergeld, 1979). Many food items which are not problematic when ingested can cause reaction in neurons when directly applied (Mattes, 1983). There is a need to determine actual levels of synthetic food additives in the central nervous system. Shaywitz, Goldenring, & Wool (1978) investigated the effects of food dyes on normal and "hyperactive" rat pups. The hyperactive-like condition was induced in the pups by exposing them to 6-hydroxydopamine (6-0HDA) at 5 days of age (this exposure reduces brain dopamine concentrations). Four groups of pups were tested: (1) hyperactive pups that received a mixture of coloring at different dose levels; (2) hyperactive pups that received water; (3) normal pups 106 that received food coloring; and (4) normal pups that received water. The activity levels of the hyperactive pups were significantly higher than those of the controls. Their activity levels at the highest dose (2 mg/kg) was increased even further. Similarly, normal pups that received a high dose of dyes displayed increased activity. Normal pups who received the dyes also demonstrated "markedly impaired" performance on both a T-maze and shuttle box. Other findings, however, were less clear. For example, activity levels decreased from baseline on 1 mg doses (Mattes, 1983) and although the hyperactive pups showed impaired performance on the maze and shuttle box, the food dyes had no further effect. In a later study, Goldenring, Wool, Shaywitz, Batter,

Cohen, Young, and Teicher (1980) fed 6-0HDA pups and normal pups a mixture of dyes through continuous gastric infusion of a liquid diet and measured subsequent activity and performance in avoidance learning trials. similar to the earlier study, there were two groups of 6-0HDA pups and two groups of sham-treated pups. One group of hyperactive and one group of normal pups were fed food dyes. The results showed that pups that received food dyes displayed significantly more activity than their no-dye counterparts. The 6-0HDA pups as a group were not more active than their sham-treated counterparts and both food- 107 dye groups showed similar increases in activity. Sham­ treated pups that were fed food dyes displayed significantly fewer successful avoidance trials than those that were not. The performance of the hyperactive pups was not affected by the dyes but these pups as a group showed significantly impaired performance in comparison to the sham-treated group (a possible ceiling effect). Although care was taken in calculating the 1 mgjkgjday dose of dyes to ensure that it approximated the average human intake, children's actual doses may be considerably higher. Also, since eight different colorings were contained in the dye mixture, it is not possible to determine which dyes or combinations were active. Finally, as in the above studies, the authors expressed the need to determine the destination (e.g., if dyes cross the blood­ brain barrier) and precise actions of the dyes. stress and Illness Feingold (1975a) wrote that "allergy can be likened to a volcano, remaining dormant for variable periods, or it may steam and sputter intermittently or persistently. In a similar manner, an individual may have an allergic state which is inactive. For various reasons, symptoms of allergy may suddenly develop. Like the volcano, these may be mild or severe, and intermittent or persistent. At times the volcano can react violently ••• Similarly, a quiescent or mild allergic state may become very active and 108 manifest itself as severe hives, eczema or asthma, and the reaction may be severe enough to cause shock and even death" (po 69). Clinical ecologists have also described this variability in expression and intensity of symptoms in individuals sUffering from food intolerances (e.g., Dickey, 1976a; Mandell & Scanlon, 1979; Randolph & Moss, 1980). One factor that may influence this variability is the amount of stress being experienced by a susceptible individual (Frazer, 1974; Strickland, 1982). Everyone has a stress tolerance threshold and exceeding this limit may result in symptoms in sensitive persons (Golos & Golbitz, 1979). This threshold obviously varies between individuals and is probably largely determined by heredity. Eating foods for which there is a susceptibility stresses the system and any additional stresses (e.g., emotional) may be the "straw that broke the camel's back," so to speak, for it taxes the body beyond its ability to adapt and symptoms become manifest. It is, of course, also possible that the food may be the "last straw" in one who is already very stressed. To date, there have been no studies on the effects of stress in reactions to foods. The following is an overview of the general model of stress and the research that provides evidence of a relationship between stress and illness. Selye's (1936, 1946, 1955, 1976) conceptualization of the stress response was eagerly 109 adopted in the 1950's and early 1960's in both clinical and experimental fields. At the core of his theory is the idea that in addition to the stimulus-specific reactions to stressors, a nonspecific pituitary-adrenal response is also involved (to be described later). For example, being struck in the eye (stressor) results in both site specific consequences including edema, trauma to various components of the eye (e.g., lens, cornea, etc.), and a nonspecific, systemic, neuroendocrinologic reaction. He claimed that this nonspecific reaction occurs in response to all stressors. He proposed that the common response is the basis for the similar complaints (i.e., fatigue, loss of appetite, headache, aching muscles and joints, insomnia, lethargy, disturbances in temperature regulation) expressed by individuals experiencing very different stressors. In Selye's model called the General Adaptation Syndrome (GAS), the body'S response to continuing stressors occurs in three phases: the alarm reaction, the stage of resistance, and the stage of exhaustion. He maintained that it was via this process that unabated stress results in physical pathologies (Le., the "disease of adaptation") and eventually death. As a consequence of the widespread acceptance of this model, much of practice and research assumed a physiologic definition of stress and focused on investigating the conditions which trigger this 110 nonspecific, all-or-none, unitary response (Appley & Trumbull, 1986a). In the mid-1960's, Selye's model began to be seriously questioned. Many held that evidence was indicating that responses to stressors were neither unitary nor all-or­ nothing (Appley & Trumbull, 1967; Lazarus, 1966; McGrath, 1970; Spielberger & Sarason, 1977). It became increasingly clear that the model, particularly the earlier formulation, was too simplistic and inflexible and did not accurately reflect what was being increasingly realized to be a very complex process involving interactions between many factors. New insights in regard to the extensiveness and complexity of the stress process resulted in the more sophisticated transactional models (e.g., Lazarus, 1966; Lehman, 1972; Leventhal, 1970). Responses to stressors in these models were viewed as a function of interactions between the whole organism and the environment thereby acknowledging the importance of psychological and situational variables. These models more satisfactorily handled a number of issues including individual differences in regard to how stressors' are perceived, reactions to stressors, ability to cope, and susceptibility or predisposition to potentially stressful stimuli. Current theories have both extended and redefined these models (e.g., Baum, Singer, and Baum, 1981; Chalmers, 111 1981; Cox & Mackay, 1981; Veno & Davidson, 1978) and there is general agreement that responses to (potential) stressors are a function of interactions between three parallel systems: physiological, psychological, and social (TrumbUll & Appley, 1986). Each of these systems consists of a number of subsystems. The physiological .system, for example, includes sUbsystems such as respiratory, digestive, nervous, and glandular processes. The psychological system consists of subsystems such as memory, perception, intelligence, emotion, and different "needs". The social system include subsystems of culture, values, SES, support networks (e.g., family, peer group, etc.). Obviously, genetic and developmental factors are important in each of these systems. The degree of stressfulness of a particular situation and the responses elicited are a function of the interactions between these three systems. In an excellent review of the current perspectives of stress, Appley and Trumbull (1986b) summarized by saying that "stress is now recognized as involving the totality of an individual's transactions with his or her environment; and importantly, such transactions must be understood both in their context and over time. In short, the dynamics of stress, as (is) now beginning to be comprehended, has become virtually as encompassing as the dynamics of life" (p. 310). Therefore, currently most investigators seem to hold a "relational, interactional, or transactional view of 112 stress, to describe stress as a process rather than a state or outcome, (and) to acknowledge its mUltilevel, mUltitemporal nature ••• " (p. 310). The effects of physical and psychological stress on hormonal functioning in animals and humans have been demonstrated and described by many (e.g., Bassett & cairncorss, 1975, 1976; Borysenko & Borysenko, 1982; Gibbs, 1986; Mason, 1975; Riley, 1981; Selye, 1976; Swenson & Vogel, 1983; Tache, DuRuisseau, Tache, Selye, and Collu, 1976; Yuwiler, 1976). Selye's theory, which expanded on Cannon's (1926, 1935) early work on homeostasis, was based on a nonspecific neuroendocrinologic mechanism (as mentioned earlier) and comprises the core of a widely accepted model. Briefly, the emotional properties of potential stressors are assessed by the cerebral cortex and this information is then communicated to the limbic system which relays it to the hypothalamus. If the stimuli have been interpreted as being emotionally-charged, the hypothalamus releases a biochemical called the corticotropin-releasing factor (CRF). Physical stress, via less known central nervous system mechanisms also cause the release of the CRF. This neuropeptide disperses to parts of the central nervous system and to the anterior pituitary gland causing the release of several hormones including adrenocorticotrophic hormone (ACTH). ACTH then acts on the adrenal cortex to 113 release adrenal cortical hormones or adrenocorticosteroids. Simultaneously, hypothalamic neurons increase the activity of the sympathetic nervous system. This increased activity causes the adrenal medulla which is directly regulated by sympathetic innervation, to release catecholamines (i.e., epinephrine or adrenaline and norepinephrine or noradrenaline). This catecholamine release then triggers the release of other hormones from this gland. The result of this process is heightened alertness and an energizing, mobilizing effect. Increased blood levels of the adrenocortical hormones then reduce CRF releases which in turn decreases the release of ACTH. In toto, the above is a basic description of the homeostatic process. The endocrine and autonomic nervous systems function interdependently in self-regulation and the maintenance of homeostasis. There is also considerable evidence indicating that physical and psychological stress also affects immunologic functioning in animals and humans (Ader, 1981; Amkraut & Solomon, 1975; Borysenko & Borysenko, 1982; Dorian & Garfinkel, 1987; Jemmott & Locke, 1984; Khalsa, 1985; Locke, 1982; Rogers, Dubey, and Reich, 1979). As is well-known, the major function of the immune system is the "maintenance of the integrity of the organism in relation to foreign substances such as bacteria, viruses, tissue grafts and organ transplantations, and 114 neoplasia" (Stein & schleifer, 1985, p. 97). Normally it accurately and expeditiously identifies and destroys "nonself" elements or antigens with minimal disruption to homeostasis of the "self". A brief description of the immunologic system as appeared in Calabrese, Kling, and Gold (1987) and Schleifer, Scott, stein, and Keller (1986) is presented below followed by a review of some of the stress-related literature. The immune apparatus can be divided into cell-mediated and humoral immunity sUbsystems. The primary components of each are the lymphocytes or white blood cells. Cell­ mediated immunity involves primarily T lymphocytes (i.e., cytotoxic T lymphocytes, helper T cells, and suppression T cells) and humoral immunity involves primarily B lYmphocytes. In the immunologic process, lymphocytes are drawn to antigens that they have been genetically programmed to recognize. The antigens then bind to the surface of the cells thereby prompting lymphocyte cell division and Ultimately antigen destruction. Reactions to antigens in subsequent exposures are more rapid and extensive due to quick differentiation of cells which, during the first encounter, did not differentiate but remained inactive and became "memory" cells. In the humoral immunity process, upon penetration of an antigen, B lymphocytes are sensitized by signals from 115 macrophages and T helper cells and proliferate and differentiate into plasma cells. These plasma cells then produce antigen-specific immunoglobulins or antibodies (i.e., IgG, IgM, IgA, IgE, IgD). IgG, the major circulating antibody (enters most tissues) coats microorganisms such as bacteria (i.e., opsonization) which then enable macrophages and neutrophils to recognize and destroy them. IgM is found in the vascular system and is capable of direct destruction of antigens. IgA is concentrated in body fluids such as saliva and tears and protect the mucosal entrances of the gastrointestinal and respiratory tracts by forming barriers rather than directly destroying antigens. IgE is the reaginic antibody that binds to allergens and then to the surface of most cells and basophils. These cells respond by releasing the mediators (e.g., histamine, kinins, leukotrienes) which cause immediate hypersenstive (allergic) reactions. The function of IgD which is present on B cell surfaces is to activate lymphocytes via a mechanism that is presently unclear. Immune processes involving B lymphocytes provide protection against infections by encapsulated bacteria (e.g., streptococci, pneumococci), neutralize toxins produced by bacteria, prevent viral reinfections, and are responsible for immediate allergic reactions such as asthma and anaphylaxis. 116 In cell-mediated immunity, T cells are directly involved in a cell-antigen interaction. This contrasts with the immune function of B cells which is primarily secretory (i.e., immunoglobulin production). T lymphocytes are sensitive to specific antigens and are located in the lymph nodes. Helper T lymphocytes mediate delayed-type hypersensitive reactions and complement the functioning of macrophage and other T and B cells. Cytotoxic T cells upon contact with an antigen release certain biochemicals (e.g., lymphokines) which both promote the function of other cells such as macrophages and after sensitization, act directly on antigens. Suppressor T cells block the production of plasma cells and the activity of helper T cells thereby serving as an important counterregulatory modulator of the immunologic process. In addition to mediating delayed allergic reactions, T cells (particularly cytotoxic T cells) protect against viral, fungal, and bacterial infection, act to reject transplants, and combat tumor growth. Along with the different types of T cells, a sUbpopulation of lymphocytes called natural killer (NK) or null cells are also actively involved in cell-mediated immune processes. These NK cells, like cytotoxic T cells, attack and destroy virally infected and tumor cells. However, unlike cytotoxic T cells which require sensitization or prior antigen interaction, these cells 117 spontaneously recognize and attack (hence the name, NK cells). A third group of white blood cells, phagocytes (polymorphonuclear and mononuclear), also have crucial functions in cell-mediated immunity. Polymorphonulcear phagocytes which include basophils, eosinophils, and neutrophils, circulate and are capable of recognizing and digesting foreign antigens. These cells also cause inflammatory and allergic reactions. The mononuclear subgroup is composed of monocytes and macrophages. Macrophages destroy pathogens, assist helper T cells, and contribute to T lymphocyte activation. Monocytes are premacrophagic circulating cells that become macrophages when they leave the vascular system and become immobilized. Due to increased understanding of the total immunologic process it is generally acknowledged that the division of immune system into humoral and cell-mediated subsystems is an oversimplification. There are complex interrelationships between components of the subsystems (e.g., functions of the helper T cells and suppressor T cells). Other aspects of functional interdependence will undoubtedly be discovered with advances in immunologic techniques and methods. Detailed treatments of the components and functioning of the immune system can be found in Bellanti (1985) and Sell (1987). 118 Most studies conducted on the effects of stress on immunologic functioning in animals and humans have provided evidence that stress has immunosuppressive consequences. For example, from animal literature, Gisler 91974) reported that restraint or crowding resulted in decreased antibody responses in mice. Monjan and Collector (1977) exposed rats to daily auditory stressors and found that the responses by lymphocytes to a B cell mitogen (i.e., substances derived from extracts of plant seeds with known ability to stimulate mitosis in lymphocytes) were suppressed. In a more recent study, Laudenslager, Ryan, Drugan, Hyson, and Maier (1983) subjected different groups of rats to avoidable shock and unavoidable shock. Compared to a "home cage" control group that received no shock, the unavoidable shock group showed suppressed lymphocyte proliferation in response to two types of mitogens as measured in an in vitro technique. There were no such effects in the avoidable shock group. There have also been two interesting studies conducted with monkeys. Laudenslager, Capitanio, and Reite (1985) studied the effects of early maternal or peer separation on lymphocyte proliferation in Macaque monkeys. In tests with three mitogens, responses of the animals who had experienced early separation were suppressed in comparison to the responses of a control group that had not had early separation experiences. As described by Dorian and 119 Garfinkel (1987), these investigators also studied the lymphoproliferation responses of infant monkeys who were separated from their mothers. The results showed suppressed responses during the two-week separation period and a return to normal (pre-separation) levels upon being reunited. In an earlier study, Reite, Harbeck, and Hoffman (1981) separated two Macaque monkey peers who had been raised together and found depressed lymphocyte stimulation to two mitogens during the. latter part of the 11-day separation and early reunion period. There have been a few reports of stressors having enhancing or no effects. Monjan and Collector (1977), for example, reported that with prolonged exposure to aUditory stress in mice, suppression is followed by enhanced proliferation. AlSO, as cited in Udelman and Udelman (1983), Gisler, Bussard, Macie, and Hess (1971) found that stress at 6, 16 or 24 hour intervals before stimulation of lymphoid cells resulted in suppressed immune activity. However, some responses after a lS-minute interval were normal or even enhanced. Overall, however, there is overwhelming evidence of immunosuppressive effects. Extensive reviews of the animal literature are presented in Borysenko and Borysenko (1982) and Rogers, Dubey, and Reich (1979). In research with humans, Bartrop, Luckhurst, Lazarus, Kiloh, and Penny (1977) studied the numbers and function of 120 T and B cells and hormone concentrations in 26 bereaved spouses. Blood samples were first drawn between the first and third week after the spouse's death and again 6 weeks later. They found that although T cell numbers were unchanged, T cell function (i.e., proliferation response to mitogens) was significantly suppressed. B cell numbers and functions were not affected. Schleifer, Keller, camerino, Thornton, and stein (1983) investigated total lymphocyte and T and B cell counts and mitogen reactions in 15 men prior and sUbsequent to the death of their wives from breast carcinoma. They found a highly significant suppression effect which was detectable as early as a month after bereavement. Total lymphocyte and Band T cell counts were not affected. The authors concluded that the results demonstrated that "the suppression of mitogen-induced lymphocyte stimulation in widowers is a direct consequence of the bereavement event" and suggested that the suppressed responses "are due either to altered numbers of sUbpopulations or to a functional defect in lymphocyte responsivity" (p. 376). Another study that examined the effects of bereavement on lymphocyte reactivity was conducted by Linn, Linn, and Jensen (1984). They measured depressive symptoms in 49 men who had recently experienced a family death or serious family illness and in 49 who had not. The results revealed that the lymphocytes of men with higher depression scores 121 in both groups were less responsive to a mitogen and allogeneic cells (i.e., lymphocytes from another individual). Many studies have also been conducted on the immunologic effects on exam stress. Dorian, Garfinkel, Brown, Shore, Gladman, and Keystone (1982) examined a group of psychiatry residents who were in the process of taking their oral fellowship exam and a matched group comprised of psychiatrists and residents who were not taking the exam. In comparison to the non-exam group, the exam group had higher lymphocyte and Band T cell counts, lower lymphocyte responses to two mitogens, and suppressed antibody synthesis two weeks prior to the exam. Two weeks after the exam, the only difference between groups was a higher responsiveness to one of the mitogens in the exam group. Kiecolt-Glaser, Garner, Speicher, Penn, Holliday, and Glaser (1984) assessed the effects of final exams on the immune responses of 75 first-year medical students. Blood samples were drawn one month before exams and on the first day of the exam period. The results showed that natural killer cell activity decreased significantly in the second sample. Total plasma IgA also decreased significantly but they were no significant changes in plasma IgG and IgM, C­ reactive protein, and salivary IgA. In a more recent study, Kiecolt-Glaser, Glaser, strain, stout, Tarr, Holliday, and Speicher (1986) examined 122 cellular immunity in 34 medical students. Similar to the earlier study, blood samples were obtained a month before exams and on the second day of the exam period. They found significant declines in the percentages of helper/inducer cells and helper/suppressor cell ratios in the second sampling. They claimed that these findings have important implications, for according to Reinherz and Scholossman (1980), these cells are responsible for important regulatory processes and interference may then, of course, result in serious immunodeficiency. Among other functions, the helper/inducer cells are involved in B cell synthesis of immunoglobulin and induce interactions between T cells and macrophages. Halvorsen and Vassend (1987) studied 12 undergraduate psychology students attending the University of Oslo. Blood samples were drawn 6 weeks before the exam (Phase I), 1 day before the first or second exam (Phase II), and 12-14 days after the exam (Phase III). A control group of 11 students not taking the exam participated in Phase II. The percentage of circulating monocytes (i.e., large white blood cells) increased and the percentage of CD4 and CD8 (i.e., types of lYmphocytes) decreased in Phase II in the exam group. The mitotic response of T cells to antigens, mitogens, and allogeneic cells decreased from Phase I to Phase III. 123 Finally, Workman and La Via (1987) looked at the effects of exam stress on T cell proliferation in 15 medical students and 15 controls. They found that differences in proliferation between pre- and postexam samplings were significantly different for the two groups with the exam group experiencing greater decreases. They found, too, that the proliferation levels in the exam group were significantly depressed for 4 weeks but returned to normal at 6 weeks. These results, then, corroborate the findings of earlier studies. In sum, these studies conducted on animals and humans very consistently demonstrated immunosuppressive effects. There are several mechanisms by which stress may exert an influence on immunologic functioning. As discussed earlier, many have investigated the effects of stress on the neuroendocrine system and reported definite hormonal responses to a variety of stressors. For example, studies cited earlier have detected increased. release of adrenocortical steroids. Glucocorticoids have positive functions in that they accelerate glucose metabolism which promotes mobilization of physical resources thereby enhancing the potential for adapting to stressors. However, evidence indicates that corticosteroids also impede the inflammatory processes by which tissue damage is controlled, decrease lymphocyte proliferation, and inhibit lYmphocyte and macrophage functioning (Claman, 1972; Fauci, 124 1978; Riley, 1981). Furthermore, according to Borysenko and Borysenko (1982), chronic exposure to corticosteroids can result in atrophy of lymphoid tissues and constant stimulation can result in hypertrophy of the adrenal glands. In short, the deleterious effects of these hormones is transient or even permanent immunosuppression and reduced immunocompetence. It was discussed earlier, too, how perceived stressors can also directly excite portions of the brain stem and induce sYmpathetic arousal. This activity then results in increased norepinephrine release from adrenergic nerve endings and also stimulates epinephrine release by the adrenal medulla. Epinephrine inhibits the function of mature lYmphocytes, other leukocytes, macrophages, basophils, mast cells, neutrophils, and eosinophils (Borysenko & Borysenko, 1982). This is, therefore, another process by which stress affects the immunologic functioning. Schleifer, Scott, Stein, and Keller (1986) also drew attention to the evidence of a direct CNS - endocrine link. They cited physiological research that have detected nerve endings in immunocompetent tissues (i.e., thymus, spleen, and lYmPh nodes). This indicates that the CNS may be able to directly affect the development and activation of immunoactive cells. 125 Finally, with improvements in techniques for detecting and measuring hormones, other hormones associated with stress have been identified. Among these are renin, glucagon, vasopressin, and thyroxin which when released, increase blood catecholamine levels (Borysenko & Borysenko, 1982). More thorough discussions of these mediating mechanisms appeared in stein, Keller, and Schleifer (1985) and Jemmott and Locke (1984). Due to its influential role in crucial physiological functions it is no surprise that stress has been implicated in the development and course of many diseases. Evidence of this association between stress and illness is provided in both animal and human models. Chang and Rasmussen (1965) inoculated mice with vesicular stomatitis virus and reported that there was increased susceptibility in mice who were subjected to aUditory stress for 6 days prior to exposure. An increased susceptibility to polyoma (tumor) virus was also reported. Riley (1975) placed groups of mice infected with the virus that causes mammary gland cancer in moderate and low stress quarters. In the low stress housing the mice were protected from distressing stimuli from other mice and the environment (e.g., pheromones, distress signals) which usually occur in standard (moderate stress) quarters. The results were dramatic in that 92% of the moderate stress as 126 compared to 7% of the low stress group had developed cancer at 400 days of age. Sklar and Anisman (1979) also studied the effects of tumor development and growth in mice. All mice were injected with equal numbers of cancer cells and divided among three conditions. In the first condition, the mice were exposed to electric shock with no means of escape. In the second condition mice were also sUbjected to shock but were able to escape by jumping over a barrier. In the third condition, the mice received no shock. As expected, tumors grew largest and most rapidly, and resulted in earliest death in mice in the inescapable shock condition. Interestingly, tumor development in mice in the escapable shock condition was no different from those in the no shock condition. In a recent review article, Dorian and Garfinkel (1987) cited many other animal studies which demonstrated associations between various stressors and increased susceptibility to tumor growth and metastases (i.e., spreading) and viruses (e.g., Amkraut & Solomon, 1972; Edwards & Dean, 1977; Hamilton, 1974; Joasoo & McKenzie, 1976; Peters, 1975; Peters & Kelly, 1977). There have also been many human studies and as with the animal studies, most have been focused on illnesses with immunologic bases. In an ambitious undertaking to 127 shed light on the relationship between infectious illness and various "host, agent and environmental variables", Meyer and Haggerty (1962) studied 16 lower-middle class families (100 sUbjects) for 12 months. Throat cultures and other laboratory measures were obtained regularly and information on chronic and acute family stress were also collected (data inclUding age, sex, contact with infected persons, allergic history, family size, type of housing, and changes in weather were also compiled). The authors reported that comparisons of two-week periods before and after acquisitions or illnesses revealed that "streptococcal acquisitions and illnesses, as well as non­ streptococcal respiratory infections were about four times as likely to be preceded as followed by acute stress" (p.

543) • Jacobs, Spilken, and Norman (1969) studied 29 male college students who had been diagnosed as suffering from an upper respiratory infection and 29 who had not. Their data showed that in comparison to the controls, those who had been ill reported higher incidence of major stressful life events inclUding identity crisis and also, as would then be expected, higher levels of negative emotions inclUding anxiety and depression. Kasl, Evans, and Niederman (1979) tested a class of 1,400 West Point Academy cadets for the presence (immune) or absence (susceptible) of the antibody to Epstein-Barr 128 virus (virus responsible for infectious mononucleosis). Among the susceptible cadets, those who experienced clinical level symptoms were much more likely to be performing poorly academically, very intent on a military career, and have fathers who were classified as overachievers. McClelland, Alexander, and Marks (1982) conducted a study on the relationship between stress and respiratory illness in prison inmates. Consistent with other findings, prisoners who appeared to be experiencing more stress, based on reports of worries, aggravations, and obstacles, suffered more severe illnesses than those who were not. In an earlier study, McClelland, Floor, Davidson, and Saron (1980) found that college males who were high in the need for power, high in inhibition, and high in power stress (i.e., experiencing life stresses such as being threatened by physical assault that are related to the power motive) reported more frequent and severe illnesses than non-HHH students. Cohen-Cole, Cogen, stevens, Kirk, Gaitan, Hain, and Freeman (1981) investigated the relationship between life changes and acute necrotizing ulcerative gingivitis or "trench mouth" (caused by a usually non-pathogenic bacteria normally found in the mouth). As compared to control SUbjects matched for age, sex, and dental hygiene, the afflicted SUbjects reported more negative life events over 129 the preceding year and more anxiety, distress, and depression. In a review of the literature on this disease, Johnson and Engel (1986) concluded that stress appears to play a role in its development by influencing catecholamine and cortisol levels. Baker and Brewerton (1981) interviewed females with rheumatoid arthritis seen within one year after the onset of sYmptoms at Westminster Hospital in London and matched age controls. Inquiries were made about important life events during the year preceding the onset of sYmptoms, relationships, emplOYment and financial security. Fifteen patients as compared to 8 controls reported stressful life events in the year prior to the onset with the events occurring for most less than three months before expression of sYmptoms. The authors concluded that their findings suggest that emotional stress preceding symptomology is a precipitating factor in this disease. An earlier study by Meyerowitx, Jacox, and Hess (1969) also reported psychological stress prior to sYmptomology in MZ twins discordant for this ailment. In a recent review, however, Koehler (1985) summarized that results of studies in this area have been mixed and more research is necessary. This review is a small sampling of a vast body of literature with emphases on the classic and more recent studies. stress has been implicated in a number of other diseases. Laudenslager and Reiter (1984) claimed that 130 separations and losses are believed to either cause or significantly aggravate over 20 different illnesses. Research has provided evidence of significant association between stress and these additional conditions: cancer (Matje, 1984; Sklar & Anisman, 1981), skin disease (Teshima, KUbo, Kihara, Imada, Nagata, Ago, & Ikemi, 1982), heart disease (Khalsa, 1985), herpes simplex virus (Luborshy, Mintz, Brightman, & Katcher, 1976), ulcerative colitis (Weiner, 1977), depression (Leff, Roatch, & Bunney, 1970; Seligman, 1975), and schizophrenic episodes (Brown, 1972). In addition, Livingston (1988) recently reviewed the research relevant to AIDS and emphasized how stress may contribute to immunocompetence and, therefore, susceptibility. In sum, although many of the studies in this area (especially those with humans) lacked adequate experimental rigor, a model of neuroendocrinologic responses to stress has been developed and is generally accepted. There is also firm evidence that immunologic functioning is influenced by these systemic processes. Finally, due largely to the better designs of the animal models, the role of stress in the predisposition, onset, and course of various disease have been rather well established. As noted earlier, veteran researchers such as strickland (1982) have in the past hypothesized about its importance but the role of stress in reactions to food has 131 not been studied. Due to the clear evidence of its immunologic impact, there are grounds to believe that IgE mediated reactions to food can be affected by stress. It should be noted, too, that with increased sophistication in studying immunologic processes, reactions that are currently considered to be intolerances may be discovered to be in fact hypersensitivities. Although the association between stress and other types of reactions to food is much less clear, there are indications that a relationship may exist. That is, it has been reported that stress also plays an important role in a number of illnesses with no demonstrated immunologic components. Furthermore, in many of the diet studies reviewed above, those experiencing adverse symptoms were in the midst of what most would acknowledge to be very stressful circumstances (e.g., hyperactives' distressing social milieu, penal environment of incarcerated juvenile delinquents, etc.). It is, therefore, important to begin investigating the effects of stress on the development and course of the different reactions to food. Hypotheses Based on the research reviewed above, the following six hypotheses were proposed. First, eliminating the foods that the sUbjects were sensitive or "addicted" to would initially impair their 132 functioning but a return to baseline levels would eventually occur with extended deprivation. Second, exposure to offending foods after two weeks of abstinence would negatively impact behavior. Third, the addition of stressful circumstances during active challenges and the first days of an elimination diet would have exacerbating effects. Fourth, when not challenged with offending foods, deterioration in behavior would only occur under stressful conditions. Fifth, in comparison to baseline levels, the most pronounced adverse effects of the elimination diet and offending foods would occur under stressful conditions. Sixth, when not challenged with offending foods, the deterioration in behavior reSUlting from the negative effects of stress would be identical to that occurring during Baseline. 133 CHAPTER 2 METHOD Design Most would agree that reactions to food (i.e., allergy and most types of intolerances) are very individualized phenomena. SYmptoms are a result of complex interactions between individual susceptibility (i.e., genetic endowment, state of health, etc.) and environment (e.g., Bock, 1980; Coca, 1942; Crook, 1975; Gontzes & Bahna, 1987; Mandell & Scanlon, 1979; May, 1984; Pearson & McKee, 1985; Randolph, 1976b, c, 1977; Randolph & Moss, 1980; Rinkel, Randolph, & Zeller, 1951; Rowe & Rowe, 1972; Walker, 1977). Strickland (1982) wrote that one of the problems that is encountered in attempts at investigating and understanding the role of foods and chemicals in behavior is lithe enormous range of individual differences in reactions. The internal and biological and physiological make-up of the individual including the presence of hormones, enzYmes, protein, and other intra and intercellular biochemicals, the psychological make-up of the person, inclUding his or her cognitive understanding and interpretation of bodily reactions and concomitant self-perceptions, the environment - both long and short term - of the individual and conditions under which he or she may ingest or inhale food and chemical sUbstances, the nature of the substances themselves and their interaction with chemicals already 134 present in the body, all playa part in individual responding. These phenomena, which can be so variable cross time, present an incredibly complicated and puzzling picture for the clinician who is attempting to identify the etiology and maintenance of a particular sYmptom. Moreover, these alarm reactions may occur independently of or in interaction with other psychological difficulties leading to a complexity of permutations extraordinarily difficult to understand and assess" (p. 41). Therefore, (1) individuals may exhibit similar or very different sYmptoms in response to the same food, (2) some individuals may react to a host of foods whereas others may react to just a few, (3) similar sYmptoms may vary in regard to intensity, frequency, and duration, (4) there may be dose­ dependent intra and interindividual differences, (5) an individual's reaction to a particular food may be mild on one occasion and severe on another, and so on. This wide range of individual responses is quite likely one of the reasons for the mixed results (e.g., subgroups of responders) obtained in the studies discussed earlier (e.g., Conners et al., 1976; Harley et al., 1978). Given these considerations, it was felt that single sUbject formats with mUltiple sUbjects were most appropriate. Many in the past have used and/or have urged using this type of design (O'Banion & Greenberg, 1982; 135 Rippers, 1983; Spring, Vermeerach, Blunden, & Sterling, 1978; Taylor, 1984; Weiss et al., 1980). SUbjects

The subjects were selected from the juvenile inmate population at Hawaii Youth Correctional Facility. The study was conducted at the facility for several important reasons: 1) control over the sUbjects' total diets, 2) Schoenthaler's (1982, 1983a, b, c, d, e) provocative early work, and 3) the procedures developed may be utilized by the institution (i.e., immediate, practical application of the results). As an initial step in identifying wards who appeared to possess food sensitivities, the personal and health records of all those with sentences which exceeded six months were reviewed. Administration and health personnel also were interviewed to screen Ior possible reactors. Two teenage males, ages 15 and 17 years, were identified in this manner. They and their parents then were interviewed to obtain more information on the boys' and families' adverse reactions to foods. A medical history questionnaire adapted from Mandell and Scanlon (1979) and Randolph (1976c) was used (see Appendices A and B) • suspicious eating habits and patterns (i.e., cravings, very frequent consumption, etc.) and histories of adverse reactions to particular foods were verified for both boys. 136 They were otherwise physically healthy with no mental disorders.

The sUbjects were told that this was another diet study and would involve taking many different vitamins in capsules (much like those in power-lifting training programs) on a couple of mornings each week and being tested each morning with different measures. They also were told that a few foods would be temporarily eliminated from their diets because these may diminish or counteract the effects of the vitamins. They believed that the purpose of the study was to determine if the vitamins influenced their performances on the daily tests. The fact that this study followed closely on the heels of a

"harmless" Hawaii state Senate mandated nutrition study

(the facility diet was simply revised to be more nutritionally balanced) helped to alleviate suspiciousness and apprehension about participating. Both boys eagerly agreed to be paid sUbjects and signed consent forms. They commented that this would break the monotony of the facility routine and also allow them to earn some money which they could use upon their release.

The parents were provided an accurate written summary of the purpose and procedure of the study. After lengthy discussions of the details and their roles and responsibilities in this project, they also signed consent 137 forms. Both sets of parents clearly understood the importance of keeping their sons blind to the purpose of the study. Moreover, since they only saw their sons for a few hours on Sundays, maintaining blind conditions was not a problem.

Apparatus

Reaction Time. A four-key serial reaction time apparatus developed by PSY-MED Associates was used to obtain the primary data. It consisted of response keys, a Commodore 64 computer, two Commodore monitors, and two Commodore 1541 disk drives. The stimulus, one of four arrows, was presented on the sUbject's monitor (Model 1701) in either of two squares (5.5 x 5.5 cm). A left square corresponded to the left hand and a right square to the right hand. The signals were white presented against a dark background. stimuli were presented randomly with never more than three consecutive displays of the same arrow. The monitor was mounted on a typewriter stand (88.5 x 49.5 x 67 cm) and positioned directly before the sUbject at a distance of about 60 cm, 15 cm below eye level. The response keys (2 x .4 cm) were Cherry E21 microswitches housed in two rectangular plastic boxes (11 x 6 x 3 cm), one for each hand with two' switches per box. The depression arms protruded .5 cm from the surface and were mounted 2 cm apart with one switch on each box positioned 138 slightly in front of the other to comfortably accommodate the sUbject's slightly separated index and middle fingers. These arms required a mere .2 cm deflection to activate. The microswitches were connected to the Commodore 64 computer which is equipped with three real-time clocks (resolutions of +/- 1 msec) to enable on-line data collection. The computer presented the stimuli and recorded the following data for each response on a 13.3 cm floppy disk: a) response latencies - reaction and dwell times, b) elapsed time, and c) accuracy - stimulus presented, response struck, and the error status. A circuit diagram of an earlier (but similar) version of this apparatus is presented by Vercruyssen (1984). The two disk drives conveniently enabled the recording of data without switching between program and data disks after each series of trials. The second monitor (Model

1802) permitted the Experimenter to view the stimuli during the trials. An on-off switch in the line of the sUbject's monitor enabled him to cut input to this monitor after the trials so that only he would receive the results that were presented after each series of trials. Blood Pressure/Pulse. The sUbject's blood pressure and pulse rate were obtained each day with a Prestige digital electronic monitor (Model DS-115). The unit was quick and simple to use. Briefly, as the sUbjects sat with their arms comfortably resting at heart level, the adjustable 139 cuff was wrapped loosely around their bare left arms (approximately 3 cm space between the arm and cuff). It was positioned about 4 cm above the elbow with the sensor in the cuff directly over the artery. The cuff was then slowly inflated with the squeeze bulb. When the unit sounded (about 200 mm Hg) the auto deflation valve was activated and the unit slowly released the pressure. At the correct pressure, blood pressure and pulse rate readings were simultaneously displayed. Detailed instruction on the correct use of this sensitive instrument was provided by the Distributor. The range of this model was 0 to 300 mm Hg cuff pressure and 40 to 150 pulse rate. The Manufacturer reported measurement accuracy +/-5% of readings for pulse rate. Time Estimation. A Sportline All Sport 210 digital stop watch (.01 sec resolution) was used in the time estimation tasks. The watch was lightweight, easy to operate, and fit comfortably in the palm of the hand. The same button was used to start and stop the watch and distinct beeps were emitted each time it was depressed. setting The study was conducted in a private office located in a far corner of the SUbjects' cottage (i.e., dorm, dining, recreation complex). The 8' x 10' room was lit by two 48" fluorescent lights and contained two louvered windows which 140 were covered by light curtains. This room was situated away from the dining and dorm areas and off-limits to the population. It was a comfortably lit and well-ventilated environment. The remote location of this room and solid concrete construction also ensured that it was quiet and distraction-free. Most testing was done in the early morning before the days' activities began while the other wards were still asleep.

Foods and Placebo

Based on previous studies, it was expected that the offending foods would likely include peanuts and other nuts, eggs, cow's milk, and soy (Atkins, 1986; Bock, 1986; Buckley & Metcalf, 1983; Mandell & Scanlon, 1979; Moneret­ Vautrin, 1986; Randolph & Moss, 1980; Rowe & Rowe, 1972; Sampson, 1983; Taylor, Bush & Busse, 1986). Surprisingly, only soy was implicated and there was evidence of only two food intolerances for each sUbject. There was strong indications from the medical histories that SUbject 1 was sensitive to coconut and soy and SUbject 2 to chocolate and shrimp.

Subject 1 reporte~ craving coconut since early childhood because "it tasted so good" and made him "feel good" or energized. In infancy, he was fed soy formula instead of mother's milk and insisted that he needed to use very liberal amounts of soy sauce with all his meals. He 141 reported, too, that on a couple of occasions he became nauseous and got severe headaches after eating several large handfuls of boiled soy beans. SUbject 2 indicated that whenever he ate shrimp he experienced stomach cramps, hives and became "angry." His mother too, recalled experiencing extreme cutaneous reactions to shrimp. This sUbject also reported that he craved chocolate and would eat as much of it as often as possible. He said that he continued to do this despite his knowing that when he overindulged, he'd feel "hyperactive", irritable, and have difficulty sleeping. The cocoa and soy in pure, powdered form were obtained from a local supermarket. The shrimp was not available in powdered form so unseasoned, dried miniature shrimps were slowly dehydrated more completely in an oven and ground with a mortar and pestle. Unsweetened finely grated coconut from a health food store was similarly slowly dried with low heat and powdered. When purchased, these foods were almost completely desiccated so only minimal additional drying was necessary. These foods were packed into capsules by the Straub Hospital pharmacy. The #1 size gray capsules were supplied by Dura Pharmaceuticals of San Diego, California. The capsules on the average totally dissolved within 3 minutes in lukewarm water. 142 As suggested by experienced allergists (e.g., Klein, Miller, ziering, 1984; Mansfield, 1984), glucose was used as the placebo. The sUbjects were tested for reactions to this substance during a single-blind phase which preceded the experimental conditions. This, of course, also was administered in the gray capsules. To minimize the intake of this sugar and avoid possible energy boosts, each capsule during the experimental phase contained only about 250 mg. The sUbjects, therefore, only ingested a total of

about 6.25 gm a day. This amount of glucose contains 26 calories which is roughly equivalent to that of 2 or 3 Lifesavers candies. Three colleagues who were not involved in this study were asked to guess the contents after smelling and handling each type of capsule. Even after very careful inspection, none were able to detect any differences between the capsules or correctly identifying the contents. They were not asked to test for differences in taste because the sUbjects were not allowed to chew or dissolve the capsules in their mouths during the study. Procedure Dependent Measures. The primary experimental tasks were two different four-choice reaction time tests. The stimulus in each was an arrow (one of four, pointing either left or right) presented on the Commodore 1701 color monitor. One test, Serial Choice Reaction Time (SCRT), 143 consisted of 150 trials with a response-stimulus internal of 0 seconds. This meant that stimulus presentations followed immediately after each response. The other test, Variable Choice Reaction Time (VCRT), consisted of 100 trials with response-stimulus intervals of 0, 1, 2, 3, or 5 seconds occurring in random order. stimuli in this test were presented at these intervals in random order. With this format, unlike that of the SCRT task, the sUbjects needed to maintain their concentration during periods of no stimuli and were unable to either predict the presentation of the next stimulus or build a rhythm of responding. The task in each of these tests was to respond as quickly and accurately as possible to the stimuli. The sUbjects placed the index and middle fingers of each hand over the four microswitches. Correct responses were depressing and releasing the switch under the left middle finger when an arrow pointing left appeared in the left box on the monitor display, using the left index finger for an arrow pointing right in the same box, using the right index finger when an arrow pointing left appeared in the right box, and using the right middle finger for an arrow pointing right in that box. In addition to the reaction time tests, there also was a time estimation task. The sUbjects estimated a period of 10 seconds using the digital sportswatch. Five trials were conducted each day. The sUbjects operated the watch 144 themselves, starting it when ready and stopping it when they thought 10 seconds had elapsed. After each trial they handed it to the Experimenter face down, without looking at the time. This estimate was recorded and the watch reset and handed back. There was no communication during these trials. On each day of this study, blood pressure and pUlse readings were obtained upon arising from sleep and again at least 20 minutes later. This was done because changes in pulse rates (Coca, 1953; corwin, Hamburger & Dukes-Dobos, 1961; Crayton, stone & Stein, 1981; Mandell & Scanlon, 1979; Randolph & Moss, 1980; Rowe & Rowe, 1972) and blood pressure (Randolph & Moss, 1980; Rowe & Rowe, 1972) have been reported to be indicators of intolerance. The final dependent variable was the amount of fidgeting displayed during each reaction time task session. This was defined as the number of minutes that the SUbjects spent tapping their feet or "bouncing"their legs. These were specific and unmistakable behaviors. In situ observer re~iability checks were conducted at least once a week. For each check, twenty 20-second intervals were selected from the days' approximately 14 minutes of reaction time testing and compared. An interval was scored as one of agreement if both the Experimenter and Observer had recorded that the behavior(s) had occurred for the same number of seconds. Reliability was determined by 145 calculating the degree of correspondence in the 20 intervals. Reliability over the course of the study ranged from 96% to 100%. SUbject Expectations and Guesses. Similar to King (1981), as a check, sUbjects' expectations were obtained before each challenge and placebo administration. A 5­ point scale ranging from "certain I will react" to "certain that I won't react" was used. In addition, after each administration, the sUbjects were asked to guess what type of capsules they had just taken and to indicate their degree of certainty on a similar 5-point scale. Stress. As discussed above, there is evidence that stress may play an important role in the development and course of adverse reactions to foods. To examine this possibility, one set of reaction time tasks (SCRT and VCRT) each day was conducted in stressful circumstances. As in other studies (e.g., Cox, 1980), during this condition sUbjects had opportunities to earn an attractive amount of money if they matched or bettered the previous days' reaction task performances. In addition, sUbjects were told before the stress trials that this set of scores would be compared to those of other groups of youngsters. These included younger and same-age boys and girls from various high schools and college students. These trials were treated as a very competitive game and emotion-charged words were used as often as possible. For example, in 146 comparing the day's scores to those of the previous day to determine if they would receive the money for the day the Experimenter would use "won" or "lost." with minor variations, the following instructions were given in the stress condition: "In this round of testing, you can win the prize money if you are as fast as or faster, than yesterday. Keep in mind, too, that your scores will be compared to those of high school students who are your age, younger than you, and college students." other potential stressors were also considered. For example, alternatives such as being able to earn privileges or points were discussed with the Facility administration and sUbjects. It became clear very early that opportunity to earn and win money would be the most effective of these because cash was the most valued. It was immediately quite obvious, too, that the boys were very competitive and "macho" and knowledge that their performances were going to be compared to others that they felt a need to be better than (i.e., girls or younger students) or disdained (i.e., college students) made it imperative to be "fastest to da max." During the Baseline period, blood pressure and pulse rates were also taken after both the stress and nonstress conditions to determine if these stressors were effective. The boys also were asked for self-reports of their anxiety 147 levels in each condition. It was evident from their elevated blood pressure and pUlse readings and their body language and remarks (e.g., swearing when they "lost" and their whoops and stamping when they "won") that they were indeed experiencing very different degrees of stress in the two conditions. These measures were, thereafter, obtained at least once in all subsequent phases to verify that the stressors maintained their potency. Practice Sessions. After the suspect foods for each sUbject had been identified, they received extensive training and practiced the different tasks. During a nonconsecutive weekdays occurring over a period of 2.5 weeks, each sUbject spent approximately 3.5 hours (about 4,000 trials each) practicing the SCRT and VCRT "games." They were instructed to depress and release the switches as quickly as possible while maintaining error rates of about 5%. This practice phase ended when their performances showed no significant improvement. Reaction times leveled off after 4 days for one boy and after 6 days for the other. During the last 2 days of the a-day practice period, they spent about 20 minutes familiarizing themselves with the stopwatch. Each boy completed approximately 50 practice trials. They were taught to start the watch, mentally count to 10, then stop the watch. In these 148 sessions they worked on improving the accuracy of their mental count (e.g., pacing, etc.). Baseline/Stress Verification. A week-long Baseline period followed the 2.5 week practice phase. Data collection on all dependent me~sures began in this period. The boys were awakened at approximately 5:45 a.m. each morning. Blood pressure and pUlse rate readings were taken about 5 minutes later in the testing room. They then returned to the dormitory to clean up and dress. One sUbject reported back 20 to 25 minutes later to begin testing. The boys alternated being first tested each morning and were not allowed to eat breakfast until tested.

Testing time each day for each subject ranged between 20 and 25 minutes. On most days testing was completed by 7:05 a.m. A second reading of blood pressure and pulse rate was taken when the sUbject returned to the testing room after dressing. This was followed by the 5 self-paced time estimation trials and the SCRT and VCRT reaction time tasks which were conducted once under both stressful and nonstressful conditions. The order of both the stress and nonstress conditions and SCRT and VCRT protocols were counterbalanced over the days in this period. That is, the boys were tested first under the different conditions on 149 alternate days and within each condition, first with the different protocols on alternate days. Amount of fidgeting also was observed during all reaction time tasks. The following instructions, with slight variations to avoid monotony, were given each day: "Please keep your fingers on the switches and press and release the correct switches as fast as you can. Remember your promise to do your very best on each test. Please keep your eyes on the screen at all times during the tests and stay seated until all the tests are completed." As mentioned earlier, a second purpose of this phase was to verify that the wards were indeed experiencing more stress in the stressful condition. Blood pressure, pulse rate, and self reports of degree of anxiety also were obtained immediately after the second reaction time task in each condition. These additional readings were taken while the computer was summarizing the sUbjects' performances and transferring the data to the data disk. It did not increase the length of the' testing period. Elimination Diet. Following the Baseline period, all suspect foods were eliminated from the sUbjects' diets for 2 weeks. All foods in the biological families of these foods were also eliminated because of possible cross­ reactivity (e.g., Knicker, 1987). Although some (e.g., McCarty & Frick, 1983) recommend up to 4 weeks of 150 avoidance, many others would agree that the 2-week period was an adequate length of time (e.g., Bock, 1980; Mandell & Scanlon, 1979; Mansfield, 1987; May, 1976; Mayron, 1974; Randolph & Moss, 1980; Rinkel, Randolf & Zeller, 1951; Rowe & Rowe, 1972). Breakfast was prepared each morning in the cottage kitchen and lunch and dinner were prepared in a central kitchen and delivered to the cottage. During this period and for the rest of the study, all suspect foods were removed from the cottage kitchen (e.g., soy sauce, cocoa, etc.) and arrangements were made with the Facility Food service Manager to eliminate these foods from most of the other meals and snacks. On the occasions when the central kitchen was not able to eliminate a prohibited food, special meals were brought in for the wards. To prevent identification of the restricted foods, special meals also were brought in when. acceptable foods were served. Since unconsumed foods and waste from meals are removed from the cottage immediately after the meal and nothing could be taken from the dining area, leftovers of these prohibited meals were not a problem.

The test foods al~o were removed from the cottage Canteen where wards could spend the points that they earned for good behavior. For example, all snacks containing chocolate, peanuts, and soy and all colas were removed and replaced with more healthy snacks like plain granola and 151 natural fruit chews. The boys' families had also agreed not to bring these foods when they visited on the weekends. They were reminded weekly and the meals that they brought in were monitored. Both families were highly cooperative and meticulously adhered to the diet. In toto, cheating or any breach of the elimination diet was very unlikely. The procedure for collecting data on the dependent measures in this phase was identical to that in the Baseline period with the exception that blood pressure and pulse rates to verify stress were not obtained everyday. Verification of Intolerance. Before launching into the time-consuming and expensive double-blind challenges, each of the suspect foods was administered single-blind to confirm reactivity and determine degree of intolerance. Although the original expectation was to administer doses in the range of 10 mg to 2" gm in this phase (Buckley & Metcalfe, 1982; Butkus & Mahan, 1986; May, 1979), it was decided that since the reported reactions were not severe 4 gm doses of each test food and glucose would be safe. While maintaining the elimination diet, the different capsules were randomly administered a day apart. Each challenge consisted of 13 capSUles. The sUbjects ingested the capSUles upon arising from sleep after the blood pressure and pUlse readings. As in the earlier phases, they then returned to the dorm and the first sUbject reported back at least 20 minutes later to begin testing. 152 The testing procedure was identical to that of the previous phase.

Double-Blind Challenges. While maintaining the elimination diet, each sUbject was tested in double-blind

format with the two different foods and placebo. The capsules were administered in random order over the next 4 weeks. The boys took active capsules on Mondays and

Tuesdays of two different weeks and placebos on these days

in two other weeks. Although full-blown reactions have been evoked with doses ranging from 20 mg to 8 gm (Bock,

Lee, Remigio & May, 1978; May, 1976; Sampson, 1983), it was decided based on the mild (yet discernible) responses during the single-blind phase that the optimum test doses would be 8 gm on Mondays followed by 12 gm on Tuesdays.

Both dose levels and the placebo were administered in 25

capsules. No capsules were administered during the other days of the week. These 5-day periods (Wednesday to

Sunday) served as purging phases, and permitted the

sUbjects' responses to return to baseline levels. Data

collection continued as in the earlier phases.

Although most of the reported reactions to these types

of challenges have not been life-threatening, there have

been some alarming accounts of profound reactions. For

example, as described earlier, the sUbject in the study

conducted by Crayton, Stone, and Stein (1981) experienced

epileptic seizures when challenged with beef. In another 153 instance, 4 children who suffered from atopic eczema went into anaphylactic shock when offending foods were introduced after a period of avoidance (David, 1984).

Arrangements were, therefore, made for a nurse to be on alert during the challenges. She was supplied with three different medications provided by George Ewing, M.D., an experienced allergist. She was familiar with their application and was prepared to summon Dr. Ewing in the event of a critical reaction. 154 CHAPTER 3

RESULTS

Test for Autocorrelation

The first step in analyzing the data from most single sUbject designs is testing for autocorrelation. Data gathered in a time series often are so related that predictions of future occurrences of an event are possible based only on knowledge of past occurrences and also may show patterns of change which are independent of any experimental manipulation (Bloom & Fischer, 1982). This likelihood of the data being confounded by serial dependency raises the possibility of erroneous results when analysis is conducted with statistical tests that assume independence of observations (Gottman & Glass, 1978).

Although autocorrelation may sometime be detectable with visual inspection because of clear increasing or decreasing trends in the baseline period, it's more commonly not obvious and more formal methods are necessary (Bloom &

Fischer, 1982). The procedure recommended by these authors

involves calculating the degree of correlation between the baseline observations (based on a lag of one) and comparing this coefficient (rk) to 2/~n, where n is equal to the number of baseline points. A rk greater than the quotient of 2/~n is interpreted as being significantly different

from zero and indication of autocorrelation. An rk that is

less than the quotient is taken as indication of 155 independence of observations. All baseline data were tested with this method and there was no evidence of autocorrelation in any of the sets of data. Sometimes visual inspection of graphed data is also sufficient to evaluate and draw conclusions from data obtained in this type of a research design. However, because the changes or discontinuities in data across phases were not of unmistakable magnitudes and there was also a possibility of complex patterns within the data, a more discerning, statistical method of analysis was needed. Shewart's (1931) procedure as presented in Bloom and Fischer (1982) was selected for these reasons and also for its simplicity and suitability when there are few baseline data points (in the present study there was only a 5-day Baseline period). In this procedure, means are calculated for the Baseline data and two standard deviation levels are constructed above and below the~e means. Statistically significant change at the .05 level is demonstrated in a particular phase if at least two consecutive observations fall beyond the two standard deviation band. Compared to visual inspection this is a conservative procedure that is obviously much more me~hodical and objective. It should be mentioned that more sophisticated (and powerful) methods of statistical analysis such as the autoregressive integrated moving average (ARIMA) model have been developed for time series data. The small number of 156 observations in the different study phases, however, made Shewart's method more appropriate for a satisfactory evaluation of the results. Error Rates The numbers of errors committed in each set of reaction time trials also were examined to determine if the results of any of the comparisons conducted to test the hypotheses may have been due to differential rates between phases. Since the reaction times calculated by the computer were based on correct responses, it was important to determine, for example, if decreases in reaction times during a particular phase were accompanied by corresponding increases in errors. Comparisons of the rates between Baseline and all subsequent phases revealed no significant differences for both sUbjects. Differences between rates obtained under stress and nonstress conditions both within phases and across phases also were not significant. The sUbjects' error rates were, then, relatively uniform between conditions and across all phases and not confounding. Measures Analysis of the data from the different measures failed to reveal consistent significant effects. The mixed findings provided minimal support for the hypotheses. The six hypotheses are restated below and the results relevant to each are presented and discussed. 157 Hypothesis 1. The first hypothesis stated that eliminating the foods that the sUbjects were sensitive or "addicted" to would initially impair their functioning but a return to usual levels would eventually occur with extended deprivation. In comparison to Baseline, slower reaction times, less accurate estimates of 10 seconds, and more time spent fidgeting in the reaction time task period were expected during the first week of the elimination diet (ELIM I) or "withdrawal" phase. Gradual improvement back to baseline levels was expected to follow during the second week of the diet (ELIM II) as the boys became completely purged of the offending foods and free of any residual effects. As predicted, subject 1 showed significant increases in VCRT reaction times in ELIM I and a return to baseline levels in ELIM II. His performance on the SCRT task, on the other hand, slowed significantly in both ELIM I and II. Elimination of the offending foods appeared to have a more enduring impact on SCRT task performance. Nevertheless, there was steady improvement during the second week. His daily SCRT (bold) and VCRT means and the corresponding two standard deviation bands are shown in Figure 1. Also contrary to expectation, there were no significant increases in the amount of time that he spent 158

380 r------:------~------....,

370 ~~··.."'········:·····:···X«·t· .... :''*\=>=«<.-1;

360

350

340

'0 (I) Ul .sw 330 :E i= r",..-----::-:::,:------+-----<\--f-l- I 320 I

I I I 310 t I <.J I ~ I c: «l t CD 300 I Cl I en C\l I I I 290 to--...... _-'~-'r---« -----I'---.-----+.L.::::e:::::::::::::::: r . I I I I I I

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 BASELINE ELIMI ELIM II PHASE/DAY

Figure 1. Reaction time during ELTh1 I and ELIM II for Subject 1 159 fidgeting in ELIM I. Surprisingly, too, as compared to Baseline, there was less within-phase variation during this week. As anticipated, his levels during ELIM II did not change from baseline. His daily levels and the two standard deviation bands are presented in Figure 2. Similar to Subject 1, SUbject 2 displayed significant increases on the SCRT task during both elimination diet phases with a general decline back toward baseline toward the end of the second week (Figure 3). His veRT performances did not change significantly in either week. The data, however, do fit the predicted pattern. There were noticeable increases from the Baseline mean (537.7 MSEC) on most days during ELIM I and on the first day of ELIM II. These changes did not achieve significance because there was substantial variation in performance on this task during Baseline and this resulted in a broad two standard deviation band. As with SUbject 1, the most clearly exhibited consequence of being without the offending foods on reaction time was the reduced pro~iciency at building a rhythm of responding when the stimuli were presented immediately after each response. Unlike Subject 1, he did display significant increases in fidgeting during both weeks of the diet (Figure 4). Data from the last day of the Baseline period (Day 5) were excluded from the analysis because he smelled of tobacco smoke that morning and it was assumed that he had taken at 160

40 .------:------.

35

30

25

'0 Ql "0c:: ~ CIS UJ 20 CD ~ 0 ~ en C\I

15

10

5

2 3 4 5 6 7 8 9 10 11 12 13 14 15 BASELINE ELiM I ELIM II PHASE/DAY

Figure 2. Fidgeting during ELIM I and ELIM II for Subject 1 161

700.------:------,

650

600

~ 550 0 1" "0c: C'll CD (J 0

450

t: o 400 en -bc: C'll CD o en C\I 350

2 3 4 5 6 7 8 9 10 11 12 13 14 15 BASELINE ELIM I ELIM II PHASE/DAY

Figure 3. Reaction time during ELIM I and ELIM II for Subject 2 162

450 r------,

400

350

300

250

200

150

100

50 t:¢o:&...--"-~«««<>_+T~ Jo» :««««:: ecc:::: : ecc: ...cu, ... I I aJ I I Cl ~.!>!o~»:-~~~~,.y.Y.oW6'.. w.w...N..~~"tW..t.. ~..oYJIl»eU...... ,(~~~H~...... H.NM'Y.oY.oW~~"""""'-+...L.U) C\l

2 3 4 5 6 7 8 9 10 11 12 13 14 15 BASELINE EUM I ELiM II PHASE/DAY

Figure 4. Fidgeting during ELIl\tl I and ELIM II for Subject 2 163 least a few puffs from a cigarette before being tested. Since nicotine acts as a stimulant, it's quite possible that the uncharacteristic, unusually elevated level was due to his smoking. The two standard deviation band was, therefore, constructed using only the data obtained on Days 1 through 4. There were dramatic increases in Elim I and although there were still increases and substantial with-in phase variability in ELIM II, there was a general decline back toward baseline levels in this second week. with the exception of the fourth day, he spent much less time fidgeting in this 'phase than in the first week. The data for both sUbjects, then, offer some support for this hypothesis. Although significance was not attained on all measures, the general pattern witnessed was deterioration during the early stages of the diet with gradual improvement in the second week. Of the effects obtained in the other phases, these were among the most striking. This was taken as strong evidence that the boys' sensitivities had been accurately identified. As shown on some of the measures, the negative effects of being without the offending foods were still quite strong at the end of the second week and a few additional days were necessary for the sUbjects' behaviors to return completely to baseline levels. 164 The daily means of the time estimation trials showed much variability within Baseline and all subsequent phases (Figures 5 and 6). There were no obvious differences between phases and the statistical analysis failed to detect any significant changes. The daily and phase standard deviations and medians also were examined because of the substantial within-phase variability but this too, failed to reveal any significant patterns of differences between phases. Interestingly, however, the daily means within each phase for SUbject 1 were quite balanced between daily means that were above and below 10.0 seconds. On the other hand, the means for SUbject 2 were very consistently above 10.0 seconds. Hypothesis 2. The second hypothesis stated that exposure to the offending foods after two weeks of abstinence would negatively impact behavior. In comparison to the Baseline phase, the deterioration on the measures expected during the first week of the elimination diet was predicted to again occur in the double blind (DB) ch~llenges. Also, since foods weren't administered in the no capsule (NO CAP) and DB placebo phases, changes from baseline levels were not expected during these weeks. For SUbject 1, the only predicted change from Baseline during the two DB active phases was increased fidgeting during the coconut challenge (Figure 7). There were no 165

I I I I I II I I I I I I I II I I ~~·.·.,.~ v..~~~~~L~~ i:lNL~'f:6...L I -e...b ...... v...... l.....-.:oe .. Mb.~1 J .L I I 11.0 1-4 I II I I I I I I I I I I I I ~ III I I I I I I I I I I I I I I I I I I II I I II I I II 10.5 - I I II 'I I I I I I I I I I I II ~ I I II I I iI I I I I 6 "0 I c: I I :. I CIS I I CD I I CI I I I en I I C\I 10·0 l- I I , I , I I I I 1 I I I I I I I I. I I I I .. \ I I I 141 I 9.5 '- I I I I I I I I I I I I I I I I II I ~I I I II I I I II II I I I I I II I I I II II I ."bW...wN...... N ..:-"...... H .....:.".Mo:.'.o·I...oMWH....Nh'~...1.y-U.Hv»y ...... "O:·....HhY.....'.oW..~..r.N...w....hY(... ~.'.hlhl ~ ~~.. ~ Y.>i>'lol I I I II I I I I I I 9.0 I I I I II I I II 1 23 4 5 6 7891011121314152122232425262728293031323334353637383940 BASELINE ELIMI ELIMII OBI NO DBII NO DBIII NO DB IV NO PLACEBO CAP I COCONUT CAP II SOY CAPIII PLACEBO CAP IV PHASE/DAY

Figure 5. Daily means of the time estimation trials for Subject 1 13.5

13.0

12.5

12.0 0' Ql S w :E i= 11.5

11.0

10.5

10.0

9.5

1 2 3 4 5 6 7 8 9 10 11 1213 1415 21 22 23 24 2526 272829 30 313233 34 35 363738 39 40 BASELINE ELI'" I ELI'" II OBI NO DB II NO DB III NO DB IV NO . SHRIMP CAP I PLACEBO CAPII PLACEBO CAP III CHOC- CAPIV OLATE PHASE/DAY

Figure 6. Daily means of the time estimation trials for Subject 2 167

140 r------,

120

100

80

60

40 ~»m'»mX'~;::::::;::::::::::mxw::~:«»>"»:"'«<_:_::::_:.l:...::;:4::1-:::""':::""'::::---1""--....,.:,,.,...---.+.::::::::;;-;roo:;":::::::;::·

2 3 4 5 26 27 31 32 BASELINE COCONUT SOY PHASE/DAY

Figure 7. Fidgeting during challenge phases for Subject 1 168 significant differences in reaction time (Figure 8). Unexpectedly, too, there was significant deterioration in SCRT reaction time performances during DB I (placebo) and NO CAP I with increased fidgeting in the NO CAP phase (Figures 9 and 10). The data for SUbject 2 are equally perplexing for there were no significant changes in reaction time performances during the two active challenges and there was a decrease in fidgeting during the shrimp challenge (Figures 11 and 12). Moreover, although there were no significant changes in reaction time during the other phases, there was a decrease and increase in fidgeting during DB III (placebo) and NO CAP IV, respectively (Figures 13 and 14). In sum, SUbject 1 only displayed increased fidgeting during one food challenge (i.e., coconut). He, furthermore, showed increases in SCRT performance and fidgeting during a placebo and no capsule phase where there should have been no changes. Similarly, SUbject 2 did not display the predicted increases in reaction time during the active challenges and, in addition, a puzzling decrease in fidgeting occurred during the shrimp challenge. Also like Subject 1, he showed significant changes in fidgeting in a placebo and no capsule phase where no effects were expected. It is difficult to assess the impact of the active challenges from these results. Although each sUbject appeared to have a limited adverse reaction to one 169

380 r------,

370

360

gt: I 350 "0 ~ ctl CD 0 C/) C\I 340

U

310 300 \ . ..

-t I I I 280 1.--:--...I...-_...1-_-l-_....J.._--..L--..L_I....-_.L...-...I...-...1-_..I-_...... 2 3 4 5 26 27 31 32 BASEUNE COCONUT SOY PHASE/DAY

Figure 8. Reaction time during challenge phases for Subject 1 170

I I I I ~o II -~·=·«,_·."="",:"",::,·t-=II,,,,·----t""""''''''''''-''l---'''+--4----r. 360 I 350

340

U' Ql UJ .§. w 330 :E i=

320

I I 310 I t II o II en I~I~ i, c:eu II "- m 300 II c II en II N II II 290 :: 'I I I I

1 2 3 4 5 21 22 23 24 25 28 29 30 33 34 35 36 37 38 39 40 BASEUNE DB I NO NO NO DB IV NO PLACEBO CAP I CAPII CAP III PLACEBO CAP IV PHASE/DAY

Figure 9. Reaction time during nonchallenge phases for Subject 1 171

100

80

-0 60 Q) S w ::E ~

40

-0 c: lU CD 20 Cl (J) C\I

1 2 3 4 5 21 22 23 24 25 28 29 30 33 34 35 36 37 38 39 40 BASELINE DB I NO NO NO DB IV NO PLACEBO CAP I CAP II CAP III PLACEBO CAP IV PHASE! DAY

Figure 10. Fidgeting during nonchallenge phases for Subject 1 172

700.....------,

650

600

t ~ 550 I -0 c: C'll CD o 0- en Q) C\l 1Il .§.. 500 w ::E i=

450

·(.:..-.:«-:.:««o»»"~.."«.:««..-':'««O'''«««O»:«««««««««<<<<<<<.:,f-----..;....-----I-+--r­

t ·W

2 3 4 5 21 22 36 37 BASELINE SHRIMP CHOCOLATE PHASE I DAY

Figure 11. Reaction time during challenge phases for Subject 2 173

175 r------,

150

125

100

(J CD .!!!. w :E i= 75

50

'.U:UUU~.lb/;6bl:lJ/;l;oI:oNoI;,J~ ",.l;lll....e._~"NN:b' __ .... -----...--""!l.!_-.... "'- - - -!-r I i:bbI:---*""""-1 "C I c: ctl 25 I m I C I en I C\I

2 3 4 5 21 22 36 37 BASELINE SHRIMP CHOCOLATE PHASE/DAY

Figure 12. Fidgeting during challenge phases for Subject 2 174 700,.------....,

I 650 I I I I I I 600 I I I I I I 550 I II g- III III :): III .§.. 500 LU :E :".: : i= III III III 450 IIII IIII

400 -,--ttt-1""'"'''' .",; ;" ,,,. """'j"""""'- IIIIII IIIIII IIIIII 350 IIIIII .;o;o;."..;."...... :o»"..;o.....»»x.«f.««Jo"~..l::::::;::::;;»»»:«~,..J.J__::_gg::_:::.lJ.-~--_.l-><_--f-.l.

1 2 3 4 5 23 24 25 26 27 28 29 30 31 32 33 34 35 38 39 40 BASELINE NO DB II NO DBIII NO NO CAP I PLACEBO CAP II PLACEBO CAP III CAP IV PHASE/DAY

Figure 13. Reaction time during nonchallenge phases for Subject 2 175

250 r------:-----:------,

200

150

100

50

....:...«.»"..:«-'.o:o'X«'.o»".«««««~J...«w"_""»".<<<<<<<<<<<<<<:<

Figure 14. Fidgeting during nonchallenge phases for Subject 2 176 food, both displayed similar effects during either placebo or no capsule phases. There was obviously no strong support for this hypothesis. verification of stress The next four hypotheses were specifically focused on the effects of stress. The data from three different measure confirmed that the sUbjects were, in fact, experiencing more stress under the stress as compared to the nonstress condition. First, in most of the phases, both subjects exhibited more fidgeting during the reaction time task periods under the stress than nonstress condition (Table 5). The only exception was during DB I for Subject 1 where the difference was a mere 1.5 seconds. Trembling, nervous tics, and inability to sit quietly without movement are listed repeatedly in the literature as strong behavioral indicators of stress (e.g., Selye, 1978). Second, on most readings, pUlse rates taken after the reaction time trials were higher following the stress condition than following the nonstress condition (Table 6). The only exceptions occurred on Days 2 and 4 for sUbject 2. Readings to verify/reverify stress were taken everyday during the Baseline period and weekly during the sUbsequent seven weeks. Evidence of a positive relationship between stress and pulse rates has been well documented (e.g., Appley & Trumbull, 1986; Selye, 1978; Zales, 1985). 177

TableS. Mean Time Spent Fidgeting Under Stress and Nonstress Conditions (sec)

SUBJECT 1 SUBJECT 2 PHASE Stress Nonstress Stress Nonstress BASELINE 10.6 9 28 10.6 ELIMI 9.4 3.8 114.6 71.2 ELIMII 6.8 5.4 51.8 35.4 SB 22 8.2 39.4 9.4 OBI 3 4.5 4 (shrimp) 1.5 NO CAP I 39.7 12 9.7 9 OBIT 58 (coconut) 32 28 24 NO CAP II 12 10.3 23 20 DB ill 26 (soy) 5.5 6.5* 2 NO CAP III 26.3 1.3 14.7* 5.3 DB IV 27* 0.5 75 (chocolate) 13.5 NO CAP IV 17* 3.6 89 26.3 x 21.5 8.0 40.3 19.0 NOTE: ELIM =Elimination, OBI =Double blind, NO CAP =No Capsule, SB =Single blind. *p<.05. 178

Table 6. PulseRates AfterStress andNonstress Conditions (bts/min) SUBJECT 1 SUBJECT 2 WEEK Stress Nonstress Stress Nonstress I Day: 1 75 61 150 106 2 115 64 105 110 3 106 91 115 69 4 67 53 56 150 5 101 89 101 91 II 150 122 150 85 ill 91 72 87 79 N 96 71 72 59 V 70 63 72 65 VI 82 69 67 60 VII 68 57 69 62 VIII 75 65 82 72 x 91.3 73.1 93.8 79 179 Third, on all days queried, both subjects reported more stress, anxiety, or nervousness after the stress than nonstress condition on the 3-point scale described earlier. As with the pulse readings, the sUbject's ratings were obtained everyday in the Baseline period and at least once a week during the following seven weeks. Under the stress condition, SUbject 1 indicated that he felt "very" stressed on 8 of the 12 days and SUbject 2 on 5 of the 12 days. On the other days they reported feeling only "a little" stressed. During the nonstress condition, Subject 1 responded "not at all" on 7 of the 12 days and SUbject 2 on 11 of the 12 days. They related being "a little" stressed on the other days. Importantly, on any given day both boys always reported feeling a higher degree of stress during the stress condition. with a rating of 3 representing high stress, the rating means in the stress condition for SUbjects 1 and 2 were 2.66 (SO = .47) and 2.42 (SO = .49), respectively. The means in the nonstress condition were 1.42 (SO = .49) and 1.08 (SO = .28), respectively. Hypothesis 3. The third hypothesis stated that the addition of stressful circumstances during active challenges and the first week of the elimination diet would have exacerbating within-phase effects. In comparison to the nonstress condition, poorer performances on the reaction time tasks and more fidgeting were expected in the stress condition during these phases. 180 As discussed above, Subject 1 did display more fidgeting during all of these phases under the stress condition. Comparisons between phase mean reaction times, however, again resulted in mixed findings (Table 7). As predicted, better performances occurred under the nonstress than stress condition during the coconut challenge on both SCRT and VCRT tasks. Contrary to expectation, though, better performances on both types of tasks occurred under the stress than nonstress condition in ELIM I and DB III (soy). SUbject 2 also engaged in more fidgeting during all phases under the stress than nonstress condition and recorded differences in reaction times that were predominantly counter to expectation. The only correctly anticipated reaction time effect was obtained on the SCRT task during the shrimp challenge (Table 8). These patterns were very surprising given what was witnessed in the lengthy practice sessions preceding the actual experimental phases. The results of the numerous practice trials were largely supportive of this hypothesis. The data discussed above generally suggest a facilitating not debilitating effect of stress on reaction time performance. Hypothesis 4. The fourth hypothesis stated that when not challenged with offending food, within-phase 181

Table 7. LowestMean Reaction Timein each Conditionfor Subject1

STRESS NONS1RESS CONDmON Scrt Vert Scrt Vert ELIMI ,,* ELIMII " DBI (placebo) " " x NO CAP I " DB II(coconut) " x x NO CAP II DB III (soy) " " NO CAP III " " x DB IV (Placebo) " x x NO CAP IV ,,* x " Unexpected lowest mean reaction time. x Expected lowest mean reaction time. * p<.05 fordifference between STRESS andNONSTRESS conditions. 182

Table 8. LowestMean ReactionTime in eachConditionfor Subject 2 STRESS NONSTRESS CONDmON Scrt Vert Scrt Vert ELIMI .J .J ELIMII .J x DBI (shrimp) .J x NO CAP I .J* .J DB II (placebo) .J* x NO CAP II .J .J* DB III(placebo) .J .J NO CAP III .J .J DB IV (chocolate) .J .J* NO CAP IV .J .J* .J Unexpected lowest mean reaction time. x Expected lowest mean reaction time. * p<.05fordifference between STRESS andNONSTRESS conditions. 183 deterioration in behavior would occur under stressful conditions. Although it was realized that there was the possibility that stress could have a facilitative effect on performance under normal conditions (i.e., without the "load" of an adverse reaction to offending foods), it was predicted that the stress induced by the opportunity to win money and the belief that their scores would be compared to those of others would be at a level that was debilitating. Slower reaction times and more fidgeting, therefore, were expected under the stress as opposed to the nonstress condition during ELIM II, NO CAP, and the DB (placebo) phases. Again, with the only exception occurring in DB I, Subject 1 displayed the predicted higher levels of fidgeting under the stress than nonstress condition in all of these phases. His reaction time data were much less consistent (Table 7). He was quicker under the nonstress than stress condition on both tasks in DB IV, only quicker under the nonstress condition on the SCRT task in three other phases, and quicker under the stress condition on both tasks in ELIM II, NO CAP I, and NO CAP II. In other words, lower reaction times on the VCRT task were obtained in most of the phases under the stress than nonstress condition and were about equally split between conditions for the SCRT task. Unexpectedly, it seemed that 184 performance on the task requiring sustained vigilance and attention was more consistently improved by stressful circumstances. Apparently, in comparison to the nonstress condition, the stress condition helped this SUbject maintain his attention and state of readiness at optimal levels when there were varying intervals of time between stimulus presentations. contrary to his consistently supportive data on fidgeting, SUbject 2 performed better on both tasks in five of the seven phases under the stress than nonstress condition (Table 8). The only correctly predicted results were obtained during ELIM II on the VCRT task and during DB II on the SCRT task. For this SUbject, then, reaction times on both tasks most often were lower under the stress than nonstress condition. The stressful circumstances enhanced his ability to respond quickly to rapidly presented stimuli as well as maintain his concentration and level of arousal during varying response-stimulus intervals. The unanticipated facilitating effect of stress is quite clear with this SUbject. Hypothesis 5. The fifth hypothesis stated that in comparison to baseline levels, the most pronounced adverse effects of the elimination diet and the offending foods would occur under stressful circumstances. As discussed earlier, there is evidence that stress may be an important, 185 if not necessary factor in the expression of symptoms. comparisons between baseline and ELIM I and the DB active challenges across stress and nonstress conditions, therefore, were expected to reveal most deterioration between behavior in the nonstress (NST) condition during Baseline and in the stress (ST) condition during these three other phases (NST vs. ST). Conversely, the least deterioration was expected in comparisons between behavior in the stress condition during Baseline and in the nonstress condition during these other phases (ST vs. NST). As shown above, in comparison to Baseline, Subject 1 displayed the predicted slower reaction times in ELIM I. These effects were evident in all comparisons with the most consistent results emerging from the ST vs. ST and ST vs. NST comparisons (Figures 15 and 16). This deterioration also was evidently stronger on the SCRT task, for significant increases on this task were obtained in all comparisons. Closer inspection of these data showed that although the differences in magnitudes of the increases between comparisons were not large, there was a pattern which ran counter to expectation. That is, the least deterioration emerged from the NST vs. ST and ST vs. ST comparisons and greatest increases emerged from the ST vs. NST comparison. These results again were indicative of a facilitating effect of stress. The comparisons between STRESS vs. STRESS STRESS VS. NONSTRESS 186 400 .------.,------..,

390

380

370

360

I I 1:: 350 I :!t I 0- "C Q) c: (/) ca ..§. CD w 340 Cl :E (J) i= if C\l I I 330 I .U:~~~l:.u.Ml~

320

1:: 310 (.) C? "C c:ca CD 300 Cl ./" (J) C\l 290 \ /

1 2 3 .4 5 6 7 8 9 10 26 27 31 32 6 7 8 9 10 26 27 31 32 BASELINE ELiM I COCO- SOY EUMI COCO- SOY NUT NUT PHASE/DAY

Figure 15. Comparison between reaction time under the stress condition during Baseline and stress (left) and nonstress (right) conditions during challenge phases for Subject 1 NONSTRESS VS. NONSTRESS NONSTRESS vs. STRESS 187

I I I I I I I I I I I I I I T t I I I I I I I 360 I I "t::: (,J I > I I "0c: I as I CD 0' Q) I 0 tJl (J) .§. 340 (\J w :E f= \!\ I I 320 I l I "t::: (,J ~ I II· I I 'f ~I II "0c: I I as CD 0 (J) II:/:~ ~lL-t-----«

1 2 3 4 5 6 7 8 9 10 26 27 31 32 6 7 8 9 1026 2731 32 BASELINE ELiM I COCO- SOY ELIMI coco- SOY NUT NUT PHASE/DAY

Figure 16. Comparison between reaction time under the nonstress condition during Baseline and nonstress (left) and stress (right) conditions during challenge phases for Subject 1 188 Baseline and the two DB food challenges, again, revealed no significant differences. The only significant change in fidgeting from baseline levels during those three phases was an increase in fidgeting during the coconut challenge (Figures 17 and 18). This finding appeared in the ST vs. ST and NST vs. ST comparisons where the strongest effects were expected to emerge. In short, in these comparisons between phases, the effects of stress were inconsistent, for it had either no influence or a facilitating effect on reaction time and a deteriorating effect on levels of fidgeting. The data for SUbject 2 were very similar to those of SUbject 1. There were reaction time increases in different comparisons in ELIM I and the effects appeared strongest on the SCRT task where both significant increases were obtained (Figures 19 and 20). The magnitudes of increases between these comparisons also showed that the largest differences on the SCRT task were obtained in the ST vs. NST comparison and the smallest in NST vs. ST comparison. This pattern which was again the reverse of what was expected, indicated that the stress condition had a facilitating effect with him, too. The data obtained in the DB chocolate phase showed no changes in reaction time. The results of the comparisons in the DB shrimp challenge were equivocal. There were the anticipated increase in 189 STRESS vs. STRESS STRESS VS. NONSTRESS 90 .------or------,

80

70

60

50

'0 OJ .!!!. w 40 :E i=

30

...... h»"..:o»"...... o»»",.:.»)»»»>'....-»»>:«o>_.....»'..f~'·...... »\-_»>w.«...... ,.,....-tw..:o.f---t»>...... -+.- 20 I I I ,

I I "0 C I I co I I m 10 C I I (J') I, ...... C\I .rV'i I I I

1 2 3 4 5 6 7 8 9 10 26 27 31 32 6 7 8 9 10 26 27 31 32 BASELINE ELiM I COCO- SOY ELIMI COCO- SOY NUT NUT PHASE/DAY

Figure 17. Comparison between fidgeting under the stress condition during Baseline and stress (left) and nonstress (right) conditions during challenge phases for Subject 1 NONSTRESS V5. NONSTRESS NONSTRESS V5. STRESS 190 90 ,....------~------~

80

70

60

50

0- Ql .!!!. w 40 ~ i=

30

20 I I I I "0 c: I ctI 10 I CD C (J) .,.-Ji C\l I o I I m»>::::::;»m>:w_:;co....::: m""'4-.L. I

1 2 3 4 5 6 7 8 9 10 26 27 31 32 6 7 8 9 10 26 27 31 32 BASELINE ELIM I COCO- SOY ELIM I coco- SOY NUT NUT PHASE/DAY

Figure 18. Comparison between fidgeting under the nonstress condition during Baseline and nonstress (left) and stress (right) conditions during challenge phases for Subject 1 STRESS vs. STRESS STRESS vs. NONSTRESS 191

650

600 ~ (J

l'"C c: CIS m \ c 550 C/) C\I

(J Ql .s-sn w 500 ::iE i= .--

450 i+---....I'--_...~_·~ ~ ~,--+-'- I II II I II II

400 ~ (J :j«·i::=::::««««>>>>>:...... «o>>"~·:««««««-.~: 0»»»»."-'..:«::::..:::::-.(; I '-+I--+-L I II II I II II

1 2 3' 4 5 6 7 8 9 10 21 22 36 37 6 7 8 9 10 21 22 36 37 BASELINE ELIMI SHRIMP CHOC- ELIMI SHRIMP CHOC- OLATE OLATE PHASE I DAY

Figure 19. Comparison between reaction time under the stress condition during Baseline and stress (left) and nonstress (right) conditions during challenge phases for Subject 2 192 NONSTRESS VS. NONSTRESS NONSTRESS VS. STRESS

I I I I I II I ;.....-.:«(>,"«~,....,(O;.:.."« ...... re..-.;.....-.:O»'~i"J;).~"«<>."'«««-:..'«««o.'w»:»>~,~((oOOOO(_(:(_::_::~+,reeeeeeeeeeeeee...... _-oxi-""""""-F""""""-+-.

650

600 t:: ~ "0c: ctl \ CD Cl 550 CJ'J C\I

U Ql 11l I .§.. 500 I w I :E I i= 1...-- I 450 I I 1...-. I t:: I o Cf "0c: 400 :--- ctl I CD I Cl CJ'J I C\I I I I 350 -"""----""-'''''''''''''''''NN---....-M·I~':_I --~

1 2 3 4 5 6 7 8 9 10 21 22 36 37 6 7 8 9 10 21 22 36 37 BASELINE ELiMI SHRIMP CHOC- ELiM I SHRIMP CHOC- OLATE OLATE PHASE! DAY

Figure 20. Comparison between reaction time under the nonstress condition during Baseline and nonstress (left) and stress (tight) conditions during challenge phases for Subject 2 193 reaction time on one task (SCRT) and an unpredicted opposing decrease on the other (VeRT) from the comparison where strong effects were not expected. Unlike with SUbject 1, the prediction of elevated levels of fidgeting was substantiated by significant increases in all comparisons between Baseline and ELIM I (Figures 21 and 22). However, in the DB shrimp challenge, significant decreases in fidgeting emerged from two comparisons which were expected to yield the largest increases. As with SUbject 1 then, these comparisons between phases also failed to reveal any coherent pattern of effects. There were the predicted deterioration in reaction time performances and increased fidgeting in ELIM I with stress having an unexpected facilitating effect on reaction time. The data from the DB phases provided little useful information, for the results of the shrimp challenge were contradictory and uninterpretable and the chocolate challenge evoked no changes. The results of the within­ phase comparisons were less ambiguous with evidence of a facilitating influence of stress but these effects were apparently not substantial enough to produce more significant between-phase differences. This is likely an indication of greater within than between-phase variability. 194 STRESS YS. STRESS STRESS YS. NONSTRESS

I I 280 I I I 260 I I I 240 I I I 220 I I I 200 I I I 180

160 '0 Ql ~ ~ 140 j:::

120

100

80

60 I I 40 1 ,.I 20 III1/ II o 1...... 1 '~''''''''''''''''''''''''''''''''''''''''''''''''''''''''H...... t''''H''''''''''''''''.v.YN''''''''''N''N''''''NN...t'''''''''''''''''''''''''''~N NNIf ,l': ...... Nw'Y\N'NN'.YNH.YNNN...... ,....N'...... ,...... N~"......

1 2 3 4 5 6 7 8 9 10 21 22 36 37 6 7 8 9 10 21 22 36 37 BASELINE ELiM I SHRIMP CHOC- ELIM I SHRIMP CHOC- OLATE OLATE PHASE/DAY

Figure 21. Comparison between fidgeting under the stress condition during Baseline and stress (left) and nonstress (right) conditions during challenge phases for Subject 2 195 NONSTRESS vs. NONSTRESS NONSTRESS vs. STRESS 300 ....------.,------.

250

200

UJ 150 :E i=

100

50

-0 ,,---''---kl»<-h-~

'NNMNWNNOO..p-t(J) C\I 1 2 3 4 5 6 7 8 9 10 21 22 36 37 6 7 8 9 10 21 22 36 37 BASELINE ELIM I SHRIMP CHOC- ELIMI SHRIMP CHOC- OLATE OLATE PHASE/DAY

Figure 22. Comparison between fidgeting under the nonstress condition during Baseline and nonstress (left) and stress (right) conditions during challenge phases for Subject 2 196 Hypothesis 6. The sixth hypothesis stated that when not challenged with offending foods, the deterioration in behavior resulting from the negative effects of stress would be identical to that occurring during Baseline. Increases in reaction times and levels of fidgeting were expected to emerge from comparisons between the nonstress condition during Baseline and the stress condition during the nonchallenge phases (NST vs. ST) and improvement was predicted in comparisons between the stress condition during Baseline and the nonstress condition during these phases (ST vs. NST). No differences were expected in the other two comparisons (ST vs. ST and NST vs. NST). For SUbject 1, some of the significant changes in reaction time in these phases offer support for this hypothesis but these are somewhat overshadowed by other unexpected results (Figures 23 and 24). That is, the predicted deterioration emerged from the NST vs. ST comparisons in only ELIM II, DB I, and NO CAP I, improvement from the ST vs. NST comparison in only NO CAP

II~ and no change from either or both of the other two comparisons in only five phases. Moreover, increases emerged from the ST vs. NST comparison in ELIM II and NO CAP I where there should have been improvement. His data on fidgeting were also mixed (Figures 25 and 26). There were the anticipated increases from the NST vs. ST comparison in NO CAP I and III but also increases from STRESS V5, STRESS STRESS V5, NONSTRESS 400 I I i I IIIII IIIIIII I I I 380 ~ I 1\ III I I IIIIII IIIIII I II II II IIIIII t::: :ro "0 c ~ o en ~ ~ ~,-- Lt~."M.,.~».L ';';".~'" l'»>:<~:<";';';'_':'hW"':'M~.N.w»»:« C\I ccccccccc ..,,;. ~ C. -l._, ~_~~~--l- - 0 320h-- -TTi -rrr- t::: o I IIII I/I~ I I II II C? I III 'I II IIIIII "0 c I IIIIII IIIIII Cll m 300 f-- '\. I 1 I 1 1---\1 I 1 I 1 1 1 I \ IV o I II IIII III ILl' en I IIIIII IIIIII C\I I IIIIII IIIIII ~«~x-:-,x-»"'-';-:«-'''~''»».1:~ -..;-x·t·~<-:''-:-'''''''';-f";..;; .«~~ ~xvx ··~«YXr·;v»;-».;.;«·»:-;.rx"·;-" x 280 ;-x-x X'';-x.:.. '\. X( t ».,,·..xcc..;ox·x-.;f·;o;ccc tx "«"O«-;f x ;- xv.--:-;..-.;-:«..• m •• xv;-;*:.x·:-;-;{..;-:-M-;·f.. xc : t..· x-:v:-···..;···N»·,·:-;···»;.;..;-: ·t "'.;.;.;.;-•••-;.;- . I IIIIII IIIIII I IIIIII IIIIII I IIIIII IIIIII

« ! «, !,, I, !!, , ,I! I ,I, ,I, I , I 260 I , !,!! It! I, ,I, I I II ,I, , ! I, , I I, ,I I,, I II 1 2 3 4 5 11 12 13 14 1521 22 23 24 25 28 29 30 33 34 35 36 37 3839 40 11 12 13 14 15 21 22 23 24 25 2829 30 33 34 35 36 3738 39 40 BASELINE ELIM II DB I NO NO NO DB IV NO ELIMII OBI NO NO NO DB IV NO PLACEBO CAPI CAP II CAP III PLACEBO CAPIV PLACEBO CAP I CAP II CAP III PLACEBO CAP IV PHASE/DAY PHASE/DAY

..... Figure23. Comparison between reaction timeunder the stresscondition during Baseline and stress (left) and \0 nonstress (right) conditions during nonchallenge phases for Subject 1 -..] NONSTRESS VS. NONSTRESS NONSTRESS VS. STRESS 400 , iii i IIII I Iii II I

1 1 I 380 F· .v.v l'llllY.'. ··..···..··· l · T ..··· r· ·..·· ··..l · T· .."..,.".....r · ,· .v.v.v.v. ,., ····lv "r..··..·· ·..·· ·..r·· · , ·..f ···· ··· ···· · r·········..·····..··l······..·.

1

t ~ I I 1 I : I 1 I I I 1 (,J 360 rI I I I I I 1 1\1I 1 I I I 1 I 1 1 IIII 1 I 1 1 1 I ~ c ctl :/\ : :\:i : :\:\ \, :I:A: : :\: aJ I, " I C U 340 (J) CD C\I !~ l V: V\: :~-: \---: - : -; ~)~ \;,,_j~l~L,; ~: \.:J; -, t .l-.__lLL_ "L _,....L..JN_l_._,,, .._" __ __..L_.LJ,,,,,," .. (,J 320 .:-»:X<<.».».,"<-:I<<<<-»»'N»:. »1:-»:«7-:.'..J»x«7\"':'ccc-x:..:,' '1 "0 1 II I I 1 IV I I I 1 I I c I I 1 I I 1 I I 1 I 1 1 I ctl I~I rn 3001-- 1 1 1 1 I 1\1 1 I 1 I C "\.. (J) 280:-r-rr1i7t1"''''''''''''''!i!i'-;r- C\I I I 1 1 I I I I I I 1 1 I 1 I 1 1 1 I 1 1 1 1 1 I 1 1 I I 1 1 1 I 1 I 1 I I I

260 I,, I,! I, ,, 1 ! I II I, !! I, II I, I ! I, I I, I, I,,, ! t II I'! , !', I "1 I III !II III 1 2 3 4 5 11 12 13 1415 21 22 23 24 2528 29 30 33 34 3536 37 38 39 40 11 12 13 14 15 2122 23 24 25282930 33 34 35 36373839 40 BASELINE ELIMII DB I NO NO NO DB IV NO ELIMII DB I NO NO NO DB IV NO PLACEBO CAPJ CAPII CAPIII PLACEBO CAPIV PLACEBO CAPI CAPII CAPIII PLACEBO CAPIV PHASE/DAY PHASE/DAY .... Figure24. Comparison between reactiontime underthe nonstress condition during Baseline and nonstress (left) \0 and stress (right) conditions during nonchallenge phasesfor Subject 1 00 STRESS vs. STRESS STRESS VS. NONSTRESS 80

70

60

50

-0 40 .e(]) w ~ i= 30

20 "0 Ceu m 10 o ~i ~:\!: en 1.]/: : 1/: i v I-Ji i1 ::\ C\I 0 ~~

,,,I,!,,I I I, , ,I III I II I III I ..10 I II,, ,I! I It! I, ,I, , ,I! I! I It' I, ,I, , I I II 1 2 3 4 5 11 12 13 14 15 21 2223 24 25 2829 30 33 34 35 36 37 38 39 40 11 12 13 14 15 21 22 23 24 2528 29 30 33 34 35 36 37 38 39 40 BASEUNE ELIMII DB I NO NO NO DB IV NO ELIMII DB I NO NO NO DB IV NO PLACEBO CAP I CAP II CAPIII PLACEBO CAP IV PLACEBO CAP I CAPII CAP III PLACEBO CAP IV PHASE/DAY PHASE/ DAY

I-' Figure 25. Comparison between fidgeting under the stress condition during Baseline and stress (left) and ~ nonstress (right) conditions during nonchallenge phases for Subject 1 ~ NONSTRESS vs. NONSTRESS NONSTRESS vs. STRESS 80

70

60

50 g 40 .e. w :E i= 30

20

"0 C ell 10 III Cl WI III I I en ,I~/" ,V, "\. C\I ' .....1 I 1...... 1 I " - 1«"1 I I I I 0 I I 1 I I 1...... 1 I I 1 I I I ,wfM~w»=;< ~""''''''"·"f"'.,"""'>+<.,m~ _""~-.~=>= «.~'m t»"'m~ ."...,.;w,-""",l,"""

,II,,,I, ! I I, , ,I! II I I t I, ! , I -10 I II,II I III, ,I I ,I I, ,I, , ,I, , t I, ,I, II I II 1 2 3 4 5 11 12 13 14 15 21 22 23 24 25 2829 30 3334 35 36 3738 39 40 1112 1314 15212223 24 252829 30 3334 35 3637383940 BASELINE ELiM II DBI NO NO NO DB IV NO ELiM II DBI NO NO NO DB IV NO PLACEBO CAP I CAP II CAP III PLACEBO CAP IV PLACEBO CAP I CAP II CAP III PLACEBO CAP IV PHASE I DAY PHASE I DAY

l\J Figure 26. Comparison between fidgeting under the nonstress condition during Baseline and nonstress (left) and o stress (right) conditions during nonchallenge phases for Subject 1 o 201 the ST vs. ST comparisons in these phases where there should have been none. In sum, these results were much like those of the within-phase comparisons. It is difficult to specify the effects of stress (especially on reaction time) between Baseline and these phases. The reaction time data for SUbject 2 also offered little support for this hypothesis (Figures 27 and 28). The only accurately predicted outcomes were no changes in NO CAP I, DB III, and NO CAP III from both the ST vs. ST and NST vs. NST comparisons and in DB II, NO CAP II, and IV from the NST vs. NST comparison. Unlike SUbject 1, no increases were obtained from the NST vs. ST comparison and no decreases emerged from the ST vs. NST comparison. In addition, there were increases from the ST vs. NST comparison in ELIM II, NO CAP I, and NO CAP III. On the fidgeting measure, the results that were predicted were again minimized by those that were not predicted

(Figures 29 and 30). Significant increases emerged from the NST vs. ST comparison in ELIM II and NO CAP IV but si9nificant changes were also obtained in these phases from either the ST vs. ST or NST vs. NST comparisons where no changes were predicted. Furthermore, there were significant decreases or improvement in NO CAP I, DB III, and NO CAP III from the NST vs. NST comparisons. The within-phase analyses, although also mixed, provided clearer evidence of a facilitating effect of stress. As in STRESS vs. STRESS STRESS vs. NONSTRESS

800iii II III iii III I I

700 t::~~ ~::::;c: ~::~:'~'(lQo,."«lO::C ««l(o;:~~~"«<'-6«-*-,

I I~I III .III III '/. V . ~'t~--III -~III'4--~- I- II ,-- ...-.-.-•.. j 400 ~"-:A.t-:f4~---i-t",t-·++~t-"- ~L"""".w",m""""",m.m~ !~i ... C\l

300' , II, ! 1, ,, ! ,', I ,I, "I ,I I I ,I, I ,I, II I) II, ), I)) ,) ) I I 1 »1 II I I, ", I), 1 234 5 111213141523242526 272829 30 3132333435383940 11 12 13 14 15 2324 2526 27 28293031 323334 3538 39 40 BASELINE ELI'" " NO 08 " NO DB III NO NO ELI'" " NO DB " NO DB III NO NO CAPI PLACE80 CAP" PLACEBO CAPIII CAPIV CAPI PLACEBO CAP" PLACEBO CAPIII CAPIV PHASE / DAY PHASE/DAY

l\J Figure 27. Comparison between reaction time under the stress condition during Baseline and stress (left) and o nonstress (right) conditions during nonchallenge phases forSubject 2 l\J NONSTRESS VS. NONSTRESS NONSTRESS VS. STRESS

800Iii i IIII I IIIIII I

700

N<*.'l(':«-»»X':(.o.».\."

1:: I 'I' III ~'\I~I ,- I VI o VI '? III 1.,1'"' IIIIIIIIIII111\1 1 1 I, IvlII ""0 C I IIII I...... ' 'I , 1,-1 I CIS 400 CXl Cl ,_L.~l_J-ilwLL_w~~Ll C/) __ __L-lH_L_. C\I ~..; "««~;..X'l.J,.;.. ~"'W'.l »X~ ;y..xv»Xo~W,.; .,'»;w.;cc":":-M*"'''"..''XmW:-J-xi«*h.;-;.l«'I."V.WX''";rxfu'.Lmxsscex c-Jcc'''W';..''Xot;;w.t...--Xseco;.....""'"",.h«t...... w...... »... ecceN:-;.1.;-;-;..ccccccox..'xL»;-:-...... xbx-.;..·....c..-;-»:J..X...... X-:-x-:-Lx-;...;.xccc.;V; c;.);-x-»;.;v;«V»;.;.:··· I IIIIII IIIIII

300 I I,,,I I, !,, ,I, ! I I, ,I, , ! I, ,I! , ,I I,, I,, I, ,I, , ,I, ,I, II I I"I,I I, I, I 1 2 3 4 5 11 1213 1415 2324 2526 272829 30 31 32 3334 3538 39 40 11 12 1314 15 2324 252627 28293031 3233 34 35383940 BASELINE ELIM II NO DBII NO DB III NO NO ELIMII NO DBII NO DB III NO NO CAPI PLACEBO CAPII PLACEBO CAPIII CAPIV CAPI PLACEBO CAPII PLACEBO CAPIII CAPIV PHASE/DAY PHASE/DAY

l\J Figure 28. Comparison between reaction time under the nonstress condition during Baseline and nonstress (left) and o stress (right) conditions during nonchallenge phases for Subject 2 w STRESS vs. STRESS STRESS vs. NONSTRESS 200 iii iii II I IIII iii

175

150

125 g 100 ~ w :::1E i= 75

50 I

2: !tl1~l]~ I IIIIII IIIIII I IIIIII IIIIII

I, ,I, , ,I, ,I, , ,', t I, , ,I, , , I -25 I 1 I , , ,I, , !, 1 I, , ,I 1 ,', , ,I, I I IIIIIII, !, 1 2 3 4 5 11 12 13 14 15 23 24 25 26 2728 29 30 31 32 33 34 35 38 39 40 11 12 13 14 15 2324 25 26 27 28 2930 31 32 33 34 3538 39 40 BASELINE ELiM II NO DB II NO DB III NO NO ELIM II NO DB II NO DB III NO NO CAPI PLACEBO CAP II PLACEBO CAP III CAPIV CAP I PLACEBO CAP II PLACEBO CAP III CAP IV PHASE/ DAY PHASE/DAY

l\J Figure 29. Comparison between fidgeting under the stress condition during Baseline and stress (left) and o nonstress (right) conditions during nonchallenge phases for Subject 2 ol'>o NONSTRESS vs. NONSTRESS NONSTRESS vs. STRESS

200 i I IIIII II IIIIIII

175

150

125

'0 Q) .!!!. w 100 ::i: i=

75

50

"0 25 C Cll co Cl (J) 0 , II ,1" II ,Ii I ,I, ~I, , ,r...."I,-=raI, , I 1"",1" !'I. ,I, ! ." .1, , ", , ! I 1, C\I 12345111213141523242526272829303132333435383940 11 121314 152324 25262728293031 3233 34 35383940 BASELINE ELIMII NO DB II NO DB III NO NO ELIMII NO DB II NO DB III NO NO CAP I PLACEBO CAPII PLACEBO CAP III CAPIV CAP I PLACEBO CAP II PLACEBO CAP III CAP IV PHASE/DAY PHASE/DAY

l\) Figure 30. Comparison between reaction time under the nonstress condition during Baseline and nonstress (left) and a stress (right) conditions during nonchallenge phases for Subject 2 01 206 the active phases, this effect was apparently not powerful enough to produce a more definite pattern of differences between phases. Once again this indicated more substantial within-phase than between-phase effects. Blood Pressure and Pulse Rate It became obvious after the first few weeks of testing that the blood pressure readings, despite the quality of the instrument used, were of questionable accuracy, and any analysis of these would not be meaningful. At times readings fluctuated wildly· and did not appear to be consistent with the pulse rates. Furthermore, when errors on the first reading were suspected, a follow-up reading had to be done on the other arm because second measures taken on the same arm within 20 minutes of the first are usually inaccurate. The frequent need for a second reading, therefore, introduced another confounding factor. Daily differences in pulse rates obtained upon arising from sleep and those taken just prior to testing for SUbject 1 are reported in Figure 31. Not surprisingly, pulse rates on.all days increased from the first to second readings. Analysis, however, failed to identify any significant changes between phases. The increases obtained on Day 3 of ELIM I and Day 1 of ELIM II were well beyond those recorded on any of the other days but not sufficient to achieve statistical significance. The increases recorded in the subsequent phases were very similar. 207 80 I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I l- 60 I I I I I I I :4 I 1 I I I I I I I I I I I I I I I I I I I '2 __L :g - I I I I I I I II ~ I I :e. 40 - I I w I I ~ a: I I w I I C/) ...J I I ;:) a. I I ~ I I w , I I C/)-c I I w I I a: 20 l- o I I z ~ , II II L II :: 'V • '1:11\:/1 : I /1...... I II II I I I III II o l-" I I I II II I I I I II II I I II II II I I III I I I I I I I II II I I ~ hY~N .,...... y ...... J...... L...... J...... JN...... N ...... L ...... J.vNNNoYo"N'NJ...... J ...... '- IIIIIIIIII I I I I I I II I I I I I I I I II I I -20 I I I III I I II 12345678 91011121314152122232425262728293031323334353637383940 BASELINE ELiMI ELiM II OBI NO DB II NO DB III NO DB IV NO PLACEBO CAP I COCONUT CAPII SOY CAPIII PLACEBO CAP IV PHASE/DAY

Figure 31. Daily differences in pulse rates for Subject 1 208 The data for SUbject 2 are presented in Figure 32. His pulse rates also increased each day from the first to second reading. As in the analysis of the data on fidgeting, data from Day 5 were also excluded from the analysis because it was believed that he smoked a cigarette after the initial reading while dressing, before reporting back to the exam room for testing. unlike SUbject 1, he showed significant increases in ELIM I and DB IV (chocolate). Unfortunately, because of the problems encountered in obtaining accurate readings, these results must be interpreted very cautiously. SUbjects' Expectations and Guesses without exception, both boys indicated that they were "certain that I won't react" on each day that capsules were administered. They were apparently totally convinced that they were taking only vitamins which they'd learned had no immediate behavioral effects and functioned primarily to keep them in good general health or at times helped them to get over a cold more quickly. Their guesses as to whether they had taken vitamins or a placebo on the capsule days were invariably "I don't know." They made comments such as the "pills all the same," "no can taste any difference," and "pills no affect me anyway." Even though they were aware of the placebo, both boys commented at different times during the study that they thought they were probably taking vitamins on all 209

70 ,-----:------::------:---:------:--:---:------,

60

50

'2 :g ~ w ~ 40 a: w en ...J ::) a.. ~ w ~ 30 w a: o ~

20

10 "C c: ~ Cl CJ) C\I

0""'" W.N.W·o 1234567 8 91011121314152122232425262728293031323334353637383940 BASELINE ELIMI ELiM II OBI NO DB II NO DB III NO DB IV NO SHRIMP CAPI PLACEBO CAP II PLACEBO CAP III CHOCo CAP IV OLATE PHASE/DAY

Figure 32. Daily differences in pulse rates for Subject 2 210 the capsule days because it would be "waste time" to give them an inert placebo. Both sUbjects, then, suspected that they were ingesting vitamins on all of the capsule days and were sure that the capsules would have no effects other than keeping them healthy and possibly preventing them from "catching flu." Thus, these results confirm that the subjects were blind in regard to the challenges and that their expectations were not confounding. 211 CHAPTER 4 DISCUSSION Review As reported, the data did not provide strong support for the six hypotheses. The most solid support was obtained for the first hypothesis which predicted deterioration on the different measures during ELIM I as the sUbjects experienced "withdrawal" and gradual improvement back to baseline levels during ELIM II as they became free of the offending foods. Subject 1 showed the predicted pattern in VCRT performances but increased SCRT times in both phases. His SCRT performances during ELIM II, however, did improve steadily back toward baseline rates as anticipated. Similarly, SUbject 2 displayed significant increases in SCRT times during both phases with noticeable improvement during ELIM II. His levels of fidgeting were also somewhat supportive of this hypothesis, for although he showed significant increases in both phases, there was, again, a discernible decrease from ELIM I during ELIM II. For Subject 1, VCRT performances and levels of fidgeting were not affected in either phase. Following the elimination diet phases, it was expected that the SUbjects would experience negative effects when the offending foods were introduced. The second hypothesis, therefore, predicted deterioration on the different measures in the two food phases and no changes in both the NO CAP and placebo phases where there were no 212 active challenges. Disappointingly, the only change from Baseline during the food challenges were increased fidgeting during the coconut challenge for Subject 1 and a puzzling decrease in fidgeting during the shrimp challenge for SUbject 2. Moreover, both sUbjects displayed changes in the other phases where effects should not have occurred. sUbject 1 showed slower SCRT performances during a placebo and NO CAP phase with also higher levels of fidgeting in this NO CAP phase. SUbject 2 displayed a decrease in fidgeting during a placebo phase and an increase during a NO CAP phase. These data, then, indicated limited reactions to one food for each sUbject but this was unfortunately obscured by similar effects in phases where changes should not have occurred. The other four hypotheses were focused on the effects of stress with the third predicting that when measures during ELIM I and the food chal,lenges were obtained once under stress and once under nonstress conditions each day, exacerbating effects would emerge during the stress condition. It was found that under the stress condition, both sUbjects did engage in more fidgeting in all of these phases, SUbject 1 also. performed more poorly on both types of reaction time tasks, and SUbject 2 had slower SCRT times during the shrimp challenge. Contrary to expectation, SUbject 1 was quicker on both tasks under the stress than nonstress condition in ELIM I and during the soy challenge. 213 Similarly, Subject 2 recorded lower reaction times on both tasks during ELIM I and the chocolate phase under the stress condition. These results more strongly indicated a facilitating than debilitating effect of stress on reaction time performance. Based on evidence of debilitating effects of stress which were obtained in pre-study trials, the fourth hypothesis also predicted poorer performances under the stress than nonstress condition in the non-active phases. The data on fidgeting for both sUbjects were, again, largely supportive of this hypothesis. Levels of fidgeting were higher in all of these phases except in a placebo phase for subject 1. The reaction time data for SUbject 1 were mixed with slower reaction times occurring under both stress and nonstress conditions over the two types of tasks during the different phases. Nevertheless, lower reaction times on the VCRT task were usually obtained under the stress condition. This indicated a facilitating effect on the task requiring sustained attention and vigilance over varying response-stimulus intervals. The reaction time results for SUbject 2 also were not uniform. His performances, however, tended to be lower over both tasks under the stress than nonstress condition. Therefore, in addition to doing better on the VCRT task, he was also quicker under stress in responding to stimuli presented in the rapid zero response-stimulus interval format. 214 Comparisons between baseline levels and the first week of the elimination diet and both food challenges were expected to reveal the most dramatic adverse effects under stressful circumstances. Hypothesis 5 predicted the highest levels of deterioration from comparisons between behavior during the nonstress condition during Baseline and during the stress conditions in these three phases. Least deterioration was expected from comparison between the stress condition during Baseline and the nonstress condition during these phases. For SUbject 1, the only significant reaction time comparisons were between Baseline and ELIM I. He showed the predicted slower reactions in ELIM I but contrary to expectation, the largest differences emerged from the comparisons where the smallest differences were predicted and vice versa. That is, the largest differences resulted from the comparisons between the stress condition during Baseline and the nonstress condition during this phase and the least deterioration resulted from comparisons between the nonstress condition during Baseline and the stress condition during this phase. His levels of fidgeting showed one significant increase during the coconut challenge and this emerged from the comparisons where the strongest effects were predicted to occur. SUbject 2 also displayed significant increases in comparisons between Baseline and ELIM I and he, too, showed 215 a pattern of differences that was the reverse of what was expected. The only other difference in reaction time occurred during the shrimp challenge where both an anticipated deterioration in seRT performance and an unexpected improvement in veRT performance emerged from a comparison where strong effects were not predicted. His data on fidgeting showed the anticipated significant increases from baseline levels in ELIM I. contrary to expectation, however, there were also significant decreases during the shrimp challenge from comparisons that should have yielded the largest increases. In short, these comparisons between phases did not reveal consistent, readily interpretable effects. stress produced different outcomes which ranged from improvement to no influence to deterioration in reaction time performances. Furthermore, it resulted in both increased and decreased levels of fidgeting. stress was expected also to produce deterioration in behavior during nonchallenge phases which was similar to that which was anticipated. during Baseline. Hypothesis 6 predicted increases in reaction times and amounts of fidgeting in comparisons between the nonstress condition (NST) during Baseline and the stress condition (ST) during these other phases and decreases in comparisons between the stress condition during Baseline and the nonstress condition in these phases. 216 There were some data from each sUbject that supported this hypothesis but these results were overshadowed by other unexpected findings. For SUbject 1, the comparisons yielded the predicted outcomes in a few of the phases. Some comparisons, however, revealed no changes where changes were predicted and effects that were the opposite of what were expected. For example, in the NST vs. ST comparisons, the anticipated deterioration in reaction time only occurred in 3 of the 7 phases. Also, reaction time increases emerged from the ST vs. NST comparison in two phases where there should have been improvement. The data on his fidgeting also provided some support for this hypothesis, for there were increases in NST vs. ST comparisons in two phases. Again, however, there were unpredicted outcomes such as increases from ST vs. ST comparisons where these should have been no change. The data for SUbject 2 offered a little stronger support in that there were no changes in reaction times in six phases from the ST vs. ST and NST vs. NST comparisons. Unfortunately, there were increases in three phases where there should have been improvement. There were also no significant improvements in levels of fidgeting in any phases from the ST vs. NST comparisons and there were three significant decreases from the NST vs. NST comparisons where there should have been no change. 217 Pulse rates for SUbject 1 increased each day from the first to the second reading but there were no significant differences in these increases between phases. Pulse rates for SUbject 2 also increased from the first to second reading on all days but unlike Subject 1, there were the predicted significant increases in ELIM I and the chocolate challenge. These results, however, must be interpreted cautiously, for the data from one day of the Baseline phase were excluded from the analysis because of suspected cigarette smoking and these readings were obtained with the same instrument used to obtain blood pressure data. These mixed findings and failure to demonstrate a clear relationship between offending foods and adverse behavioral reactions are similar to what has occurred in many previous studies in all the major research areas in this field. Although many studies have provided some evidence of associations between foods (or food SUbstances) and behavior, these findings were minimized by factors such as accompanying contradictory results or lack of corroborating data from different measures. For example, from the studies of the Feingold diet, Conners, et al. (1976) found that for most SUbjects, teachers reported significant improvement in behavior based on ratings on the Conners Questionnaire but parents did not. Also, a few of their preschool sample were rated as improved by both teachers and parents but this was not sufficient to result in 218 significance for both parent's and teacher's ratings in aggregate analysis. Rose (1978) in examining the effects of tartrazine on two 11-year old girls could similarly only report mixed evidence of adverse effects, for there were predicted changes on only two of the three behavioral measures. In reaction to the dye challenge, out-of-seat behavior increased in frequency and duration and on-task behavior decreased in duration. The third dependent variable, physical aggression, did not show the expected concomitant increase. Rowe (1988) studied the effects of this and another azo dye, carmoisine (red). Two of his sample of nine children were rated to have displayed significant behavioral reactions to the challenges including increased activity and sleep disturbances. The validity of these results, however, were somewhat diminished by generally low correlations between the ratings of the parents and teachers on the other sUbjects and on other behaviors. Levy, Dumbrell, Hobbes, Ryan, wilton, and Woodhill (1979) also challenged hyperactive elementary school children with tartrazine and found that a subsample of children who had respo~ded favorably to the Feingold diet were rated by their mothers as significantly more hyperactive during the challenges. The results of the other measures including scores on the Continuous Performance and Draw-A-Line-Slowly Tests weren't fully 219 consistent, for although some scores mirrored the ratings, others were contradictory or failed to show any differences. King's (1981) study from the Clinical Ecology literature, is another example. He used a variety of measures including self-report, signature size, pulse rate, WAIS and Digit/Symbol Substitutions test scores. The results of the double-blind food challenges showed that, compared to the placebo trials, reports of cognitive­ emotional symptoms were significantly higher and the standard deviations of the heart rate changes were significantly larger. In addition, there were also requests for relief from unpleasant symptoms (e.g., administration of oxygen) during the food trials. No corroborating changes occurred on the other measures. As in the other studies, then, there were significant indications of adverse reactions but these were not consistent across all measures. Research from the traditional approaches in the food allergy field such as that by Bock, Lee, Remegio, and May (1978) found evidence of behavioral reactions to milk, soy, and peanuts. These allergists were primarily concerned with verifying the classic. symptoms including skin rash, respiratory problems, and upset stomach but also reported behavioral aberrations as reactions to the test foods. They emphasized, however, that the behavioral symptoms were 220 never the sole reactions. It could be argued that the behavioral reactions were merely secondary symptoms or reactions to the somatic manifestations. There have also been sugar studies which yielded provocative evidence of behavioral effects but failed to demonstrate these convincingly. Rosen, Bender, Sorrel, Booth, McGrath, and Drabman (1988) tested pre-school and elementary age children in three conditions: high sugar, low sugar, and aspartame (control). Measures included simple academic tasks, associate learning, and ratings of activity level. Aggregate results showed that the girls made significantly more errors on the cognitive measures during the high sugar as compared to the low sugar condition and there were differences between age groups in their responses to the challenges. Unfortunately, none of the analyses on the individual tasks were significant so identifying the specific effects of the independent variable was impossible and interpretation of the results difficult. Schoenthaler's (1982, 1983a, b, c, d, e) series of studies in 10 juvenile correctional facilities found evidence of a link between diet and antisocial behavior. His low sugar, low refined carbohydrate diet was reported to have reduced antisocial behavior by 40% in most of these institutions. These were impressive results. However, due to the lack of an accurate measure of the actual reduction 221 in sugar and refined carbohydrate consumption and the questionable reliability of the measures of the dependent variable, the influence of this diet is uncertain. Finally, from the animal research, Shaywitz, Goldenring, and Wool (1978) found that a 2 mg/kg dose of food dye significantly increased the activity levels of hyperactive and normal mice. Normal mice also displayed deteriorations in performance on both a T-maze and Shuttle box. With 1 mg/kg doses, however, activity levels decreased from baseline levels. This inconsistent finding raised serious questions about the predicted outcome. Design The primary reason for the mixed results in many of these and the present study is very likely weaknesses in design. Although the procedure in this study was painstakingly designed based on previous research and successful clinical protocols, it unfortunately proved to be still in need of refinement. A review of the methodological inadequacies and a discussion of ways to improve the procedure follow. Baseline. The Baseline period was limited to a week because of a strict timeline imposed by the Facility. Extending this period would have enabled both obtaining more data and attaining more within-phase stability. This was especially important because the two-standard deviation bands which were used to determine the significance of 222 changes between phases were constructed from the data gathered in this period. with more data points and stability, the precision and usefulness of this method of analysis can be improved. Furthermore, most methods of analysis require that within-phase stability be established and more baseline data make it also possible to apply other statistical tests in the analysis. It is necessary in research of this type that the design be flexible enough to permit extending the Baseline period until a convincing steady state is achieved. Elimination Diet. The significant and even dramatic changes during the elimination diet were among the most unequivocal effects obtained. Based on the results of previous research (e.g., Bock, 1980; May, 1976; Randolph & Moss, 1980; Rowe & Rowe, 1972), two weeks should have been more than adequate time for behavior to return to normal levels. As was discovered, however, this was apparently too short a period, for the negative effects were still obvious in the second week. A longer period of abstinence appeared necessary for these effects to abate more completely. Again, it is essential that there be the flexibility in the design to extend those phases in which re-establishing stability is crucial. Foods. As discussed above, there were limited adverse reactions to one food for each sUbject in this study. 223 Unfortunately, these were obscured by similar and unexpected results in other phases. The test foods were selected by the detailed medical questionnaire and lengthy interviews with the sUbjects, their families, and the Facility nurse. This procedure was reported to be accurate and reliable in clinical practice (Randolph, 1976c; Randolph & Moss, 1980). The significant deterioration in behavior during the elimination diet was taken as verification of intolerance to the selected foods. Furthermore, in the subsequent single blind challenges intended as an additional screening, SUbject 1 displayed increased reaction times beyond the two standard deviation margin in response to both soy and coconut. Subject 2, likewise, showed changes in reaction times after exposure to chocolate and shrimp. These results, then, provided even more evidence of sensitivity. In light of the mixed results, however, even this method of identifying offending food appears to be inadequate. The single blind phase can be improved by extending exposure or administering additional doses over a longer period of time•. This would eliminate the possibility of one-time, accidental (false) reactions. Also, corroboration from other types of tests such as skin tests and the RAST may further increase the accuracy of the assessment. 224 All indications from earlier studies (e.g., Bock, Lee, Remegio, & May, 1978, Samson, 1983) were that the dosage levels and schedule on which the capsules were administered would be sufficient to provoke sYmptoms. Another possible explanation for the very limited responses may be that the reprieve from the offending foods during the elimination diet resulted in renewed resilience and that continuous exposure was necessary to elicit symptoms which would ordinarily appear in a weakened or more vulnerable state. It's prudent that the challenges be extended and include more prolonged and sustained exposure to graduated doses. The foods should also be administered to approximate the sUbject's routine intake. stress. The effects of the stress condition on reaction time was quite unexpected given that the sUbjects were experiencing a high degree of stress in this condition as indicated by blood pressure, pulse rate, and self report. In general, this condition appeared to improve reaction time performances. A likely reason for this unanticipated effect is the higher level of motivation which existed in this condition. The facilitating rather than debilitating effect on their performances was probably due to the strong incentive created by the opportunity to win money. The belief that their scores would be compared to groups that they felt superior to or very competitive with 225 may have been enough to induce an adequate degree of stress. This would comprise a "purer" stress condition without the possible confounding effects of a powerful reinforcer like money. Incorporating a procedure of presenting degraded stimuli, as in Vercruyssen (1989, unpublished), may also help achieve the desired effect. That is, sUbjects believing that their scores will be compared to others that they are competitive with coupled with stimuli which are difficult to accurately identify will surely produce high (or higher) levels of stress. Measures. The reaction time measure was satisfactory and will be even more useful when the different procedural weaknesses are addressed. It is a sensitive and reliable method of obtaining important data on cognition. Moreover, the sUbjects were challenged and thoroughly enjoyed the tasks. The results of the time estimation task were of little value because of the excessive variability within each phase. The procedure can be improved by requiring the sUbjects to achieve a designated high level of proficiency before the experimental trials (i.e., as was done with the reaction time tasks). Deviations from these levels in the different phases can then be examined. This procedure would render it a much more refined measure. Despite the support for using blood pressure as evidence of sensitivity (e.g., Coca, 1953; Crayton, stone, 226 & stein, 1981), it was learned that blood pressure normally fluctuates considerably for many different and oftentimes undeterminable reasons. Factoring out the "noise" to measure reactions is difficult and probably not worth the effort. If it is to be used, however, the data should be obtained with equipment such as a polygraph or EKG apparatus. These would provide readings that are superior to manual readings or those obtained from portable models. In addition, these enable continuous monitoring of blood pressure and pulse rate instead of mere "snapshot" readings. Unfortunately, social and more naturalistic measures such as the proposed classroom behavior ratings were not permitted in the present setting. These types of measures are important for they provide information on different facets of behavior and shed light on the effects of intolerances on the total functioning of the individual. These types of measures need to be included because they provide glimpses of behavior outside the confines of the testing room and beyond laboratory tasks. Previous research and clinical practice (e.g., May & Bock, 1978) have shown that most reactions to challenges of this kind are immediate, that is, occur soon after digestion. The lack of significant symptoms, however, may suggest delayed reactions. Optimally, data gathering should extend beyond just immediate readings and occur 227 continuously at predetermined intervals throughout the day and phase. In this study, measures taken every hour .until lunch (a period of approximately four hours) may have revealed delayed symptoms. Future Research Despite the fact that most studies have provided only marginal evidence of an association between food and behavior, research in this area mus~ continue for at least several reasons. First, there has been unequivocal demonstrations of behavioral reactions in a few studies (e.g., Crayton, stone, & stein, 1981; Goldman, Lerman, Contois, and Udall, 1986). Second, there has been some evidence of behavioral responses in many studies including the present study and those which were just reviewed above. Third, an adequate research design has yet to be developed and until this is achieved, conclusions can not be drawn with confidence. Since controlled investigation in this area is relatively new, every thoughtful study contributes to the kno~ledge of methodology and/or understanding of this phenomenon. The value of this study lies primarily in the former. To summarize, it was learned that, first, a flexible single sUbject or time series approach is even more of a necessity than was initially realized. Due to the very individual nature of reactions and high probability of unexpected results in research of this type, 228 these designs permit adjustments in face of the unanticipated. Second, the procedure for selecting test foods must be more thorough with an extended single blind procedure and inclusion of other types of screening tests. Third, the challenges or exposures to the foods should be more prolonged and sustained in case of cyclic or cumulative sensitivities. Fourth, some of the measures need to be refined to be more useful and measures of other behaviors need to be included. Fifth, data must be gathered regularly over longer periods of time. Sixth, the effects of stress can be studied more effectively by modifying and experimenting with different stressors. Finally, decisions to conduct future research in a correctional institution must be weighed carefully because of some major disadvantages. HYCF was initially considered to be the ideal setting because .of the control over the sUbject's total diets but due to its primary function as a youth correctional facility, it proved to be too restrictive in regard to timelines and also in areas such as the types of measures that were permissible. Although research needs to eventually proceed beyond the laboratory, at thi~ stage of study, the next series of behavioral studies should be conducted in hospitals or in­ patient clinics. In these settings with adult volunteers, informed consent, adequate testing facilities, and access to emergency medical treatment and more sophisticated 229 testing equipment are not formidable obstacles. Before moving to less clinical, "sterile" settings, more useful methodologies need to be developed and behavioral reactions must be confirmed empirically in an environment which maximizes the probability of success. There are too many substantive indications of behavioral reactions to foods for it not to be a genuine phenomenon. The task ahead is to more effectively identify the population and circumstances in which it occurs. Appendix A. Medical History Questionnaire-Part I SYMPTOMS CAROIOVASCUlAR IlEAllATOlllGIC GASTROINTESTINAL IIUSCULOSKELETAI. I£UIlOl.OllIC RESPIRATORY OTIER SkIp Sholl· P.,.. FM\IUOI D1u~ !:0n- FOOOS .pod Tid!}'- Illft. Ilah _Ino 1JIIr· Na> Vorrt· Jok1I Dopm 0..- Htal· e-- ~ EM'· Enlhu- IIc!lIng .-, CIlIdIa .- CnoT1>I IN Pain Adl. n_ 0·· .... ~ Ing IEdomal .- Ing 11- oiled alaallc T!woIt =: 01 ~- lion -. ------81_ --AppIo -.-BellI CIIlbIge COnOII Celery Chidu0 Ot~ 0>'11" POldl ".... P- P... Pear-. PInIl4lPlo

Scallop -Sh""" Soy Squid Tom3IO walnu.. N v_ W - o 231

Appendix B. Medical History Questionnaire-Part II

Yes No

1. Are there any foods youloveor crave?

2. Are there any foods youdislike or hate?

3. Are there any foods thatmake yousick?

4. Do you feel uncomfortable (notjust hungry) if you skipa meal or don't eat at theusualtime?

5. Which foods do youeat frequently or regularly (morethanonce every three or fourdays)?

Food Frequency

1. 2. 3. 4. 5. 232 REFERENCES

Aas, K. (1978). The diagnosis of hypersensitivity to

ingested foods. Reliability of skin prick testing

and radioallergosorbent test with different material.

Clinical Allergy, ~, 39.

Ader, R. (Ed.). (1981). Psychoneuroimmunology. New York: Academic Press. Alvarez, w. c. (1946). puzzling "nervous storms" due to food allergy. Gastroenterology, 7, 241-242.

Alvarez, w. c. , & Hinshaw, H. C. (1935) • Foods that commonly disagree with people. Journal of the

American Medical Association, 104, 2053-2058.

American Academy of Allergy (1981). position statements ­

controversial techniques. Journal of Allergy and

Clinical Immunology, 67, 333-338. American Dietetic Association (1985). position paper of

the American Dietetic Association on diet and

criminal behavior. Journal of the American Dietetic

Association, 85(3). Amkraut, A. A., & Solomon, G. F. (1972). Stress and murine sarcoma virus (Maloney)-induced tumors. Cancer

Research, 32, 1428-1433. Amkraut, A. A., & Solomon, G. F. (1975). From the symbolic stimulus to the pathophysiologic response: Immune

mechanisms. International Journal of Psychiatry in

Medicine, ~, 541-563. 233 Anderson, J. A. (1984). Non-immunologically-mediated food sensitivity. Nutrition Reviews, 42, 109-116.

Anderson, J. A. (1986). The establishment of common

language concerning adverse reactions to foods and

food additives. Journal of Allergy and Clinical Immunology, 78, 140-143.

Anderson, J .A. (1988). The pediatrician's guide to food allergy. Henry Ford Hospital Medical Journal, 36, 198-203. Anderson, J. A., & Sogn, D. D. (Eds.). (1984). Adverse reactions to foods. American Academy of Allergy and

Immunology and National Institute of Allergy and

Infectious Diseases. (stock No. 017-044-0045-1). Washington, DC: u.s. Government Printing Office. Appley, M. H., & Trumbull, R. (Eds.). (1967). Psychological stress: Issues in research. New York:

Appleton-Century-Crofts.

Appley, M. H., & Trumbull, R. (1986a). Development of the stress concept. In M. H. Appley & R. Trumbull (Eds.), Dynamics of stress (pp. 3-18). New York: Plenum Press. Appley, M. H., & Trumbull, R. (1986b). Dynamics of stress and its control. In M. H. Appley & R. Trumbull (Eds.), Dynamics of stress (pp. 309-327). New York: Plenum Press. 234 Atherton, D. J. (1985). Skin disorders and food allergy. Journal of the Royal Society of Medicine, 78, 7-10. Atherton, D. J., Soothill, J. F., Sewell, M., Wells, R. S., & Chilvers, C.E.D. (1978). A double-blind controlled crossover trial f antigen avoidance diet in atopic eczema. Lancet, i, 401. Atkins, F. M. (1986). A critical evaluation of clinical trials in adverse reactions to foods in adults. Journal of Allergy and Clinical Immunology, 78, 174­ 182. Atkins, F. M., steinberg, S. S., & Metcalfe, D. D. (1985a). Evaluation of immediate adverse reactions to foods in adult patients: Pt. I. Journal of Allergy and Clinical Immunology, 75, 348-355. Atkins, F. M., Steinberg, S. S., & Metcalfe, D. D. (1958b). Evaluation of immediate adverse reactions to foods in adult patients: Pt. II. Journal of Allergy and Clinical Immunology, 75, 356-363. Augustine, G. J., & Levitan, H. (1980). Neurotransmitter release from a vertebrate neuromuscular synapse affected by a food dye. Science, 207, 1489-1490. Bahna, S. L. (1987). Food sensitivity. Postgraduate Medicine, 82, 195-209.

Baker, G. H. t & Brewerton, D. A. (1981). Rheumatoid arthritis: A psychiatric assessment. British Medical Journal, 282, 2014. 235 Barkin, G. D., & McGovern, J. P. (1966). Allergy statistics. Annals of Allergy, 24, 602-609.

Bartrop, R. W., Luckhurst, E., Lazarus, L., Kiloh, L. G., & Penny, R. (1977). Depressed lymphocyte function after bereavement. Lancet, April 16, 834-836. Bassett, J. R., & Cairncross, K. O. (1975). Time course for plasma 11-hydroxycorticosteroid elevations in rats during stress. Pharmacology, Biochemistry &

Behavior, ~, 139-142. Bassett, J. R., & Cairncross, K. O. (1976). Endogenous levels of catecholamines in rat myocardium following exposure to stress. Pharmacology, Biochemistry &

Behavior, ~, 35-38.

Baum, A., Singer, J. E., & Baum, D. S. (1981). Stress and the environment. Journal of Social Issues, 37, 4-35. Behar, D., Rapoport, J. L., Adams, A. J., Berg, C. J., & Cornblath, M. (1984). Sugar challenge testing with

children considered behaviorally "sugar reactive. 1I

Nutrition and Behavior, ~, 277-288.

Bell, I. R. (1975). A kinin model of mediation for food

and chemical sensitivities: Biobehavioral

implications. Annals of Allergy, 35, 206-215.

Bellanti, J. A. (Ed.). (1985). Immunology: Basic

processes. Philadelphia: Saunders. Bender, A. E., & Matthews, D. R. (1981). Adverse reactions to foods. British Journal of Nutrition, 46, 403-407. 236 Bennett, F. C., & Sherman, R. (1983). Management of

childhood "hyperactivity" by primary care physicians.

Developmental and Behavioral Pediatrics, ~, 88-93. Bernstein, M., Day, J. H., & Welsh, A. (1982). Double­ blind food challenge in the diagnoses of food

sensitivity in the adult. Journal of Allergy and Clinical Immunology, 70, 205-210.

Bock, S. A. (1980). Food sensitivity: A critical review

and practical approach. American Journal of Diseases

of Children, 134, 973-982.

Bock, S. A. (1982). The natural history of food

sensitivity. Journal of Allergy and Clinical

Immunology, 69, 173-177.

Bock, S. A. (1986). A critical evaluation of clinical

trials in adverse reactions to foods in children.

Journal of Allergy and Clinical Immunology, 78, 165­

174.

Bock, S. A. (1987). Prospective appraisal of complaints of

adverse reactions to foods in children during the

first 3 years of life. Pediatrics, 79, 683-688. Bock, S. A., & Martin, M. (1983). The incidence of adverse

reactions to foods -A continuing study. Journal of

Allergy and Clinical· Immunology, 71, 113. Bock, S. A., Lee, W., Remigio, B. s., & May, C. D. (1978). Studies of hypersensitivity reactions to foods 237 infants and children. Journal of Allergy and

Clinical Immunology, 62, 327-334.

BoIsen, B. (1982). No agreement on diets for "hyperactive"

kids. Journal of the American Medical Association,

247, 948-956. Borysenko, M. & Borysenko, J. (1982). stress, behavior, and immunity: Animal models an mediating mechanisms.

General Hospital Psychiatry, ~, 59-67.

Breneman, J. C. (1985). Food allergy -A new science.

Journal of the Arkansas Medical Society, 81, 594-611.

Brenner, A. (1977). A study of the efficacy of the

Feingold Diet on hyperkinetic children. Clinical

Pediatrics, 16, 652-656.

Brown, G. W. (1972). Life-events and psychiatric illness:

Some thoughts on methodology and causality. Journal

of Psychosomatic Medicine, 16, 311-320.

Buckley, R. H., & Metcalfe, D. (1982). Food allergy. Journal of the American Medical Association, 248,

2627-2631-

Buisseret, P. D. (1978). Common manifestations of cow's

milk allergy in children. Lancet, Feb. 11, 304-305.

Buisseret, P. D., Youlten, L. J. F., Heinzelmann, D. I., &

Lessof, M. H. (1978). prostaglandin-synthesis

inhibitors in prophylaxis of food intolerance.

Lancet, Apr. 29, 906-908. 238 Burr, M. L., & Merrett, T. G. (1983). Food intolerance: A community survey. British Journal of Nutrition, 49, 217-219. Businco, L., Benincori, N., & Cantani, A. (1984). Epidemiology, incidence and clinical aspects of food allergy. Annals of Allergy, 53, 615-621. Businco, L., Benincori, N., & Cantani, A. (1986). The spectrum of food allergy in infancy and childhood. Annals of Allergy, 57, 213-218.

Butkus, s. N., & Mahan, L. K. (1986). Food allergies: Immunological reactions to food. Journal of the American Dietetic Association, 86, 601-608. Calabrese, J. R., Kling, M. A., & Gold, P. W. (1987). Alterations in immunocompetence during stress, bereavement, and depression: Focus on neuroendocrine regulation. The American Journal of Psychiatry, 144, 1123-1134. Campbell, M. B. (1973). Neurologic manifestations of allergic disease. Annals of Allergy, 31, 485-498. Cannon, W. B. (1926). The emergency function of the adrenal medulla in pain and the major emotions. American Journal of Physiology, 33, 356-372. Cannon, W. B. (1935). stresses and strains of homeostasis. American Journal of Medical Science, 189, 1. Cant, A. J. (1985). Food allergy in childhood. Human Nutrition: Applied Nutrition, 39A, 277-293. 239 cant, A. J., Gibson, P., & Dancy, M. (1984). Food hypersensitivity made life threatening by ingestion

of aspirin. British Medical Journal, 288, 755-756.

Chalmers, B. E. (1981). A selective review of stress: Some cognitive approaches taken a step further. Current

Psychological Reviews, 1, 325-344. Chang, s. S., & Rasmussen, A. F. (1965). Stress induced suppression of interferon production in virus­ infected mice. Nature, 205, 623. Check, w. (1985). Eat, drink, and be merry - or argue about food 'allergy.' Journal of the American

Medical Association, 250, 701-711.

Claman, N. H. (1972). Corticosteroids and lymphoid cells. New England Journal of Medicine, 287, 388-397.

Clarke, T. W. (1944). Allergy of the central nervous

system. Annals of Allergy, ~, 189-279.

Clarke, T. W. (1950). The relation of allergy to character

problems in children. Annals of Allergy, 8, 175-187.

Clein, N. W. (1954). Cow's milk allergy in infants. Pediatric Clinics of North America, Nov., 949-962.

Coca, A. F. (1942). Familial nonreaginic food allergy, 1st

ed. springfieldl. Illinois: Charles C. Thomas. Coca, A. F. (1953). Familial nonreaginic food-allergy.

springfield, Illinois: Charles C. Thomas. Cohen-Cole, S., Cogen, R., Stevens, A., Kirk, K., Gaitan, E., Hain, J., & Freeman, A. (1981). Psychosocial, 240 endocrine, and immune factors in acute necrotizing ulcerative gingivitis ('trenchmouth'). Psychosomatic Medicine, 43, 91-95. Collins-Williams, C. (1956). Incidence of milk allergy in pediatric practice. Journal of Pediatrics, 48, 39. Conners, C. K. (1969). A teacher rating scale for use in drug studies with children. American Journal of Psychiatry, 126, 884-888. Conners, C. K. (1970). Symptom patterns in hyperkinetic, neurotic, and normal children. Child Development, 41, 667-682. Conners, C. K. (1980). Food additives and hyperactive children. New York: Plenum Press. Conners, C. K., Eisenberg, L., & Barcai (1967). Effect of dextroamphetamine on children. Archives of General Psychiatry, 17, 478. Conners, C. K., Goyette, C. H., & Newman, E. B. (1980). Dose-time effect of artificial colors in hyperactive children. Journal of Learning Disabilities, 13, 48­ 52. Conners, C. K., Goyette, C. H., Southwich, D. A., Lees, J. M., & Andrulonis, P. a. (1976). Food additives and hyperkinesis: A controlled double-blind experiment. Pediatrics,. 58, 154-166. Conners, D., Wells, K., Horn, W., Blouin, A., Beerbohm, E., O'Donnell, D., Seidel, W., & Shaw, D. (1982). The 241 effects of sucrose and fructose on behavior of child psychiatric patients. Sugar Associates, Inc. Consensus Conference (1982). Defined diets and childhood hyperactivity. Journal of the American Medical Association, 248, 290-292. Cook, P. 5., & Woodhill, J. M. (1976). The Feingold dietary treatment of the hyperkinetic syndrome.

Medical Journal of Australia, ~, 85~90. Cook, P. S., & Woodhill, J. M. (1977). Food additives and

hyperactivity. Medical Journal of Australia, ~, 188­ 189. Coombs, R. R. A. (1984). Marabou symposium on food sensitivity. Nutrition Reviews, 42, 117. Coombs, R. R. A., & Gell, P. G. H. (Eds.). (1963). Clinical Aspects of Immunology. Blackwell, Oxford. Corwin, A. H., Hamburger, M., & Dukes-Dobos, F. N., (1961). Bio-assay of food allergens. Annals of Allergy, 19, 1300-1311. Cox, R. H. (1980). A comparison of cardiovascular responses in reactive and non-reactive individuals to psychological and physical challenge (Doctoral dissertation, university of Tennessee, 1980). Dissertation Abstracts International, 41, 1963. Cox, T., & Mackay, C. (1981). A transactional approach to occupational stress. In E.N. Corlett & J. Richardson 242 (Eds.), stress, work design, and productivity (pp.

91-113). New York: Wiley.

Crawford, L. V., Lieberman, P., Harfi, H. A., Hale, R.,

Nelson, H., Selner, J., Wittig, H., Postman, M., &

Zietz, A. (1976). A double-blind study of

subcutaneous food testing sponsored by the Food

Committee of the American Academy of Allergy.

Journal of Allergy and Clinical Immunology, 57, 236. Crayton, J. W., Stone, T., & stein, G. (1981). Epilepsy precipitated by food sensitivity: Report of a case

with double-blind placebo-controlled assessment.

Clinical Electroencephalography, 12, 192-198.

Crook, W. G. (1975). Food allergy - The great masquerader.

Pediatric Clinics of North America, 22, 227-238. Crook, W. G., Harrison, W. W., Crawford, S. E., & Emerson, B. S. (1961). Systemic manifestations due to

allergy. Pediatrics, 27, 790-799.

D'Asaro, B., Groesbeck, C., & Nigro, C. (1975). Journal of

Orthomolecular Psychiatry, ~, 212-222.

David, T. J. (1984). Anaphylactic shock during elimination

diets for severe atopic eczema. Archives of Disease

in Childhood, 59, 983-986.

David, T. J. (1985). The overworked or fraudulent

diagnosis of food allergy and food intolerance in

children. Journal of the Royal Society of Medicine,

78, 21-31. 243 David, T. J. (1987). Reactions to dietary tartrazine.

Archives of Disease in Childhood, 62, 119-122 •.

Davison, H. M. (1949). Cerebral allergy. Southern Medical

Journal, 42, 712-716.

Dees, S. C. (1959). Allergy to cow's milk. Pediatric

Clinics of North America, ~, 881-890.

Denman, A. M. (1980). The relevance of immunopathology to

research in schizophrenia. In G. Hemmings (Ed.), Biochemistry of schizophrenia and addiction.

Lancaster: MTP Press.

Dickey, L. D. (Ed.) (1976a). Clinical ecology.

Springfield, Illinois: Charles C. Thomas.

Dickey, L. D. (1976b). History and documentation of the

food factors of disease. Food allergy - Fact or fiction? In L. D. Dickey (Ed.), Clinical ecology (pp. 26-35). springfield, Illinois: Charles C.

Thomas.

Dickerson, J. W. T., & Pepler, F. (1980). Diet and hyperactivity. Journal of Human Nutrition, 34, 167-

174. Dorian, B., & Garfinkel, P. E. (1987). stress, immunity and illness - a review. Psychological Medicine, 17,

393-407.

Dorian, B., Garfinkel, P., Brown, G., Shore, A., Gladman, D., & Keystone, E. (1982). Aberrations in lymphocyte subpopulations and functions during psychological 244 stress. Clinical and Experimental Immunology, 50,

132-138. Edwards, E.A., & Dean, L.M. (1977). Effects of crowding on mice humoral antibody formation and protection to

lethal antigenic challenge. Psychosomatic Medicine,

39, 19-24. Egger, J., Carter, C. M., Graham, P. J., Gumley, D., & Soothill, J. F. (1985). Controlled trial of the oligoantigenic treatment in the hyperkinetic

syndrome. Lancet, March 9, 1985, 540-545. Egger, J., Carter, C. M., Wilson, J., Turner, M. W., & Soothill, J. F. (1983). Is migraine food allergy? Lancet, Oct. 15, 865-869.

Faelton, S. (1983). The allergy self-help book. Pa.:

Rodale Press.

Fand, R., & Hubbard, R. C. (197~). Acute anxiety reaction secondary to food allergy. Connecticut Medicine, 40,

188-189. Farah, D. A., Calder, I., Benson, L., & Mackenzie, J. F. (1985) • Specific food intolerance: Its place as a

cause of gastrointestinal symptoms. Gut, 26, 164­

168.

Fauci, A. S. (1978). Mechanisms of the immunosuppressive and anti-inflammatory effects of glucocortico­

sterioids. Journal of Immunopharmacology, ~, 1-25. 245 Federal Register: United states Department of Agriculture -

rules and regulations. 48:32749, 1983.

Feingold, B. F. (1968). Recognition of food additives as a

cause of symptoms of allergy. Annals of Allergy, 26,

309-313.

Feingold, B. F. (1973a). Food additives and child

development. Hospital Practice, October, 11-21.

Feingold, B. F. (1973b). Introduction to clinical allergy.

springfield, Illinois: Charles C. Thomas.

Feingold, B. F. (1975a). Why your child is hyperactive.

New York: Random House.

Feingold, B. F. (1975b). Hyperkinesis and learning

disabilities linked to artificial food flavors and

colors. American Journal of Nursing, 75, 797-803.

Feingold, B. F. (1977a). A critique of "controversial medical treatments of learning disabilities".

Academic Therapy, 13, 173-183.

Feingold, B. F. (1977b). Behavioral disturbances linked to

the ingestion of food additives. Delaware Medical

Journal, 49, 89-94.

Ferguson, H. B., stoddart,. C., & Simeon, J. G. (1986). Double-blind challenge studies of behavioral and

cognitive effects of sucrose-aspartame ingestion in

normal children. Nutrition Reviews, 44, 144-150.

Ferreira, s. H., Moncada, S., & Vane, J. R. (1971). Indomethacin and aspirin abolish prostaglandin 246 release from spleen. Nature (New Biology), 231, 237- 239.

Fink, J. (1984). Help is on the way for 35 million with

allergies. U.S. News & World Report, ~, 67. Finn, R., & Cohen, H. N. (1978). "Food allergy": Fact or fiction. Lancet, Feb. 25, 426-428. Fishbein, D. (1981). The contribution of refined carbohydrate consumption to maladaptive behavior. Unpublished doctoral dissertation. Florida state University. Ford, R. P. K., & Fergusson, D. M. (1980). Egg and cow's milk allergy in children. Archives of Disease in Childhood, 55, 608.

Forman, J. (1932). On being poisoned by harmless foods, so-called food atopy. Journal of the American

Dietetic Association, ~, 316-322.

Frazier, C.A. (1974). Coping with food allergy. New York: New York Times Book Company. Freir, S., & Kletter, B. (1970). Milk allergy in infants

and young children. Clinical Pediatrics, ~, 449. Fries, J. H. (1959). Factors influencing clinical evaluation of food allergy. Pediatric Clinics of

North America, 6, 867-879. Gallagher, J. S., Riehm, J. g., Valanis, B., & Bernstein, I. L. (1983). Cross sectional survey of the 247 incidence of adverse food symptoms in adults.

Journal of Allergy and Clinical Immunology, 71, 113.

Gerrard, J. W., Ko, C. G., Vickers, P., & Gerrard, C. D. (1976). Annals of Allergy, 36, 10-15.

Gettis, A. (1987). Serendipity and food sensitivity: A case study. Headache, 27, 73-75.

Gibbs, D. M. (1986). Vassopressin and oxytocin: Hypothalamic modulators of the stress response: A review. Psychoneurendocrinolgy, 11, 131-140.

Gisler, R. H. (1974). stress and hormonal regulation of the immune response in mice. Psychotherapy and

Psychosomatics, 23, 197-208. Gisler, H. R., Bussard, E. A., Mazie, C. J., & Hess, R. (1971). Hormonal regulation of the immune response.

1. Induction of an immune response in vitro with

lymphoid cells from mice exposed to acute systemic

stress. Cellular Immunology, ~, 634.

Glass, G. V. (1976). Primary, secondary, and meta-analysis

of research. Educational Researcher, ~, 3-8.

Glass, G. V. (1977). Integrating findings: The meta­

analysis of research. In L. Shulman (Ed.). Review

of Research in Education, ~, 351-379.

Goldenring, J. R., Wool, R. S., Shaywitz, B. A., Batter, D. K., Cohen, D. J., Young, J. G., & Teicher, M. H. (1980). Life Sciences, 27, 1897-1904. 248 Goldman, A. S., Anderson, D. W., Sellers, W. A., saperstein, S., Knicker, W. T., Halpern, S. R. , & collaborators. (1963). Milk allergy. Pediatrics, Sept., 425-443. Goldman, J.A., Lerman, R.H., contois, J.H., & Udall, J.N. (1986). Behavioral effects of sucrose on preschool children. Journal of Abnormal Child Psychology, 14, 565-577. Golos, N., & Golbitz, F. G. (1979). Coping with your allergies. New York: simon and Schuster. Gontzes, P., & Bahna, S. L. (1987). Food allergy for the primary care physician. Primary Care, 14, 547-558. Goyette, C. H., Conners, C. K., Petti, T. A., & Curtis, L. E. (1978). Effects of artificial colors on hyperkinetic children: A double-blind challenge study. Psychopharmacology Bulletin, 14, 39-40. Gray, G. E. (1986). Diet, crime and delinquency: A critique. Nutrition Reviews, 44, 89-94. Gray, G. E., & Gray, L. K. (1983). Diet and juvenile delinquency. Nutrition Today, May/June, 14-22. Gross, M. D., Tofanelli, R. A., Butzirus, S.M., & Snodgrass, E. W. (1987). The effects of diets rich in and free from additives on the behavior of children with hyperkinetic and learning disorders. Journal of the American Academy of child and Adolescent Psychiatry, 26, 53-55. 249 Hall, K. (1976). Allergy of the nervous system. Annals of Allergy, 36, 49-64. Halpern, G. M., & Scott, J. R. (1987). Non-IgE antibody mediated mechanisms in food allergy. Annals of Allergy, 58, 14-27. Halvorsen, R., & Vassend, O. (1987). Effects of examination stress on some cellular immunity functions. Journal of Psychosomatic Research, 31, 693-701- Hamilton, D. R. (1974). Immunosuppressive effects of predatory induced stress in mice with acquired immunity to hymenolepis nana. Journal of Psychosomatic Research, 18, 143-153. Hannington, E. (1980). Diet and migraine. Journal of Human Nutrition, 34, 175-180. Hannuksela, M., & Haahtela, T. (1987). Hypersensitivity reactions to food additives. Allergy, 42, 561-575. Hansel, F. K. (1933). Allergy as related to otolaryngology

and pediatrics. Journal of Pediatrics, ~, 516-522. Hanson, L. A. (1984). Marabou symposium on food sensitivity. Nutrition Reviews, 42, 117.

Harley, J. P., & Matthews, C. G. (1979). The Feingold Hypothesis: Current studies. The Journal of the Medical Society of New Jersey, 76, 127-129.

Harley, J. P., Matthews, C. G., & Eichman, P. (1978). Synthetic food colors and hyperactivity in children: 250 A double-blind challenge experiment. Pediatrics, 62, 975-983. Harley, J. P., Ray, R. S., Tomasi, L., Eichman, P. L., Matthews, C. G., Chun, R., Cleeland, C. S., & Traisman, E. (1978). Hyperkinesis and food additives: Testing the Feingold Hypothesis. Pediatrics, 61, 818-828. Hawley, C., & Buckey, R. (1974). Food dyes and hyperkinetic children. Academic Therapy, 10, 27. Heiner, D. C., Wilson, J. F., & Lahey, M. E. (1964). Sensitivity to cow's milk. Journal of the American Medical Association, 189, 117. Hippenchen, L. J. (Ed.). (1978). Ecologic-biochemical approaches to treatment of delinquents and criminals. New York: Van Nostrand Reinhold. Hippenchen, L. J. (1981). Some possible biochemical aspects of criminal behavior. International Journal

of Biosocial Research, ~, 37-42. Hoobler, B. R. (1916). Some early symptoms suggesting protein sensitization in infancy. American Journal of Diseases of Children, 12, 129-125. Hopkins, J. G., Waters, I., & Kesten, B. (1931). Elimination diets as an aid in the diagnoses and

treatment of eczema. Journal of Allergy, ~, 239-245. Hosen, H. (1989). Letters to the editor. Annals of Allergy, 61, 319. 251 Intraagency Collaborative Group on Hyperkinesis: First Report of the preliminary Findings and Recommen­ dations. U.S. Department of Health, Education and Welfare, January, 1976. Jacob, L. H., & Carron, D. B. (1987). A trial fibrillation precipitated by tyramine containing foods. British Heart Journal, 57, 205-206. Jacobs, M. A., Spilken, A., & Norman, M. (1969). Relationship of life- change, maladoptive aggression, and upper respiratory infection in male college students. Psychosomatic Medicine, 31, 31-44. Jakobsson, I., & Lindberg, T. (1983). Cow's milk proteins cause infantile colic in breast-fed infants: a double-blind crossover study. Pediatrics, 71, 268. Jemmott, J. B., & Locke, S. E. (1984). Psychosocial factors, immunologic mediation, and human susceptibility to infectious diseases: How much do we know? Psychological Bulletin, 95, 78-108. Joasoo, A., & McKenzie, J. M. (1976). stress and the immune response in rats. International Archives of Allergy and Applied Immunology, 50, 659-663. Johansson, S. G. O. (1984). Immunological mechanisms of food sensitivity. Nutrition Reviews, 42, 79-84. Johnson, B. D., & Engel, D. (1986). Acute necrotizing ulcerative gingivitis. A review of diagnosis, 252 etiology and treatment. Journal of Periodontology, 57, 141-150. Karnosh, L. J. (1944). Psychosomatic aspects of allergy. Psychiatric quarterly, 18, 618. Kasl, S. V., Evans, A. S., & Neiderman, J. C. (1979). Psychosocial risk factors in the development of infectious mononucleosis. Psychosomatic Medicine, 41, 445-466. Kavale, K. A., & Forness, S. R. (1983). Hyperactivity and diet treatment: A meta-analysis of the Feingold Hypothesis. Journal of Learning Disabilities, 16, 324-330. Khalsa, D. S. (1985). stress-related illness. Postgraduate Medicine, 78, 217-221. Kenyon, J. N. (1986). Clinical ecology. The Practitioner, 230, 1095-1103. Kettelhut, B.V., & Metcalfe, D.O. (1987). An overview of the controversy concerning questionable manifestations of food allergy. Annals of Allergy, 59, 146-149. Kiecolt-Glaser, J. K., Garner, W., speicher, C., Penn, G. M., Holliday, J., & Glaser, R. (1984). Psychosocial modifiers of immunocompetnce in medical students. Psychosomatic Medicine, 46, 7-14. Kiecolt-Glaser, J. K., Glaser, R., strain, E. C., stout, J. C., Tarr, K. L., Holliday, J. E., & Speicher, C. E. 253 (1986). Modulation of cellular immunity in medical

students. Journal of Behavioral Medicine, ~, 5-21. King, D. S. (1981). Can allergic exposure provoke psychological symptoms? A double-blind test. Biological Psychiatry, 16, 3-19. King, W.P., Fadal, R.G., Ward, W.A., Trevino, R.J., Pierce, W.B., stewart, J.A., & Boyles, J.H. (1988). Provocation-neutralization: A two-part study. Part II. Subcutaneous neutralization therapy: A multi­ center study. otolaryngology-Head and Neck Surgery, 99, 272-277. King, Q.P., RUbin, W.A., Fadal, R.G., Ward, W.A., Trevino, R.J., Pierce, W.B., Stewart, J.A., & Boyles, J.H. (1988). Provocation-neutralization: A two-part study. Part I. The intracutaneous provocative food test: A multi-center comparison study. otolaryngology-Head and Neck Surgery, 99, 261-271. Kittler, F. J., & Baldwin, D. G. (1970). The role of allergic factors in the child with minimal brain dysfunction. Annals of Allergy, 28, 203-206. Klein, G.L., Miller, M.F., & Ziering, R.W. (1984). A simplified food challenge for allergic rhinitis and atopic dermatitis. Paper presented at the annual meeting of the American Association of Clinical Immunologists and Allergists, New Orleans, LA. 254 Knicker, W.T. (1987). Immunologically mediated reactions to food: state of the art. Annals of Allergy, 59, 60-70. Koehler, T. (1985). stress and rheumatoid arthritis: A survey of empirical evidence in human and animal stUdies. Journal of Psychosomatic Research, 29, 655­ 663. Kolata, G. B. (1978). Childhood hyperactivity: A new look at treatments and causes. Science, 199, 515-517. Kwok, R. H. M. (1968). Chinese-restaurant syndrome. New England Journal of Medicine, Apr. 4, 796. Lafferman, J. A., & Silbergeld, E. K. (1979). Erythrosin B inhibits dopamine transport in rat caudate synaptosomes. science, 205, 410-412. Lancet (1979a). Food allergy. Feb. 3, 249-250.

Lancet (1979b). Feingold's reg~men for hyperkinesis. Sept. 22, 617-618. Laudenslager, M., Capitanio, J. P., & Reite, M. (1985). Possible effects of early separation experiences on subsequent immune function in adult Macaque monkeys. American Journal of psychiatry, 142, 862-864.

Laudenslager, M. L., &. Reite, M. L. (1984). Losses and separations: Immunological consequences and health implications. In P. Shaver (Ed.), Review of personality and social psychology. California: Sage Publications. 255 Laudenslager, M. L., Ryan, S. M., Drugan, R. C., Hyson, R. L., & Maier, S. F. (1983). Coping and immuno­ suppression: Inescapable but not escapable shock

suppresses lymphocyte proliferation. Science, 221,

568-570.

Lazarus, R. S. (1966). Psychological stress and the coping process. New York: McGraw-Hill.

Lee, C. H. (1961). A new test for diagnosis and treatment

of food allergies. Bachanan Co. Medical BUlletin,

25, 9. Leff, M. H., Roatch, J. F., & Bunney, w. E. (1970). Environmental factors preceding the onset of severe

depression. psychiatry, 33, 293-311.

Lehman, E. C. (1972). An empirical note on the

transactional model of psychological stress. The

Sociological Quarterly, 13, 484-495.

Lehman, C. W. (1980). A double-blind study of SUblingual

provocative food testing: A study of its efficacy.

Annals of Allergy, 45, 144-149.

Leinhas, J. L. , McCaskill, C. C., & Sampson, H. A. (1987). Food allergy challenges: Guidelines and implications.

Journal of the American Dietetic Association, 87,

604-608.

Lessof, M. H. (1983). Food intolerance and allergy -A

review. Quarterly Journal of Medicine, LII. No. 206,

111-119. 256 Lessof, M. H. (1984). Food intolerance and food aversion.

Journal of the Royal College of Physicians of London,

~, 83-123.

Lessof, M.H. (1988). Clinical reactions to food. British

Journal of Hospital Medicine, 39, 138-142. Lessof, M. H., Wraith, D. G., Merrett, T. G., Merret, J., & Buisseret, P. D. (1980). Food allergy and

intolerance in 100 patients - Local and systemic effects. Quarterly Journal of Medicine, XLIX, 259­

271.

Leventhal, H. (1970). Findings and theory in the study of

fear communications. In E. L. Berkowitz (Ed.),

Advances in experimental social psychology: Vol 5

(pp. 120-186). New York: Academic Press. Levine, M.D., & Liden, C. B. (1976). Food for inefficient thought. Pediatrics, 58, 145-148.

Levitan, H. (1977). Food, drug, and cosmetic dyes:

Biological effects related to lipid solubility.

Proceedings of the National Academy of Science, 74,

2914-2918.

Levy, F., Dumbrell, S., Hobbes, G., Ryan, M., Wilton, N., &

Woodhill, J. M. (1978). Hyperkinesis and diet.

Medical Journal of Australia, 1, 61-64.

Levy, F., & Hobbes, G. (1978). Hyperkinesis and diet: A replication study. American Journal of Psychiatry,

135, 1559-1560. 257 Lew, F. (1977). The Feingold Diet experienced. Medical

Journal of Australia, ~, 190. Linn, M. w., Linn, B. S., & Jensen, J. (1984). Stressful events, dysphoric mood, and immune responsiveness. Psychological Reports, 54, 219-222. Lipton, M. A., & Mayo, J. P. (1983). Diet and hyperkinesis - an unpdate. Journal of the American Dietetic

Association, 83, 132~134. Livingston, I. L. (1988). Co-factors, host susceptibility

and AIDS: An argument for stress. Journal of the National Medical Association, 80, 49-59. Locke, S. E. (1982). stress, adaptation, and immunity.

General Hospital psychiatry, ~ 49-58. Lockey, S. D. (1959). Allergic reactions due to FD&C yellow no. 5 tartrazine, an aniline dye used as a coloring and identifying agent in various steroids. Annals of Allergy, 17, 719-721. Lockey, S. D. (1973). Drug reactions and sUblingual testing with certified food colors. annals of Allergy, 31, 423-429. Lockey, S. D. (1975). Reactions to hidden agents in foods and drugs can be serious. Annals of Allergy, 35, 239-242. Lockey, S. D. (1977). Hypersensitivity to tartrazine (FD&C yellow no. 5) and other dyes and additives present in 258 foods and pharamceutical products. Annals of Allergy, 38, 206-210. LUborsky, L., Mintz, J., Brightman, V. J., & Katcher, A. H. (1976). Herpes simplex virus and moods: A longi­ tudinal study. Journal of Psychosomatic Research, 20, 543-548. Mandell, F. G. (1981). Dr. Mandell's allergy-free cookbook. New York: Simon and Schuster. Mandell, M. (1976). Ecologic concepts and techniques: Their impact on an allergist and his practice. In L. D. Dickey (Ed.), Clinical ecology (pp. 610-629). Springfield, Illinois: Charles C. thomas. Mandell, M., & Scanlon, L. W. (1979). 5-day allergy relief system. New York: Pocket Books. Mansfield, L.E. (1984). Protocol: Office blind food challenge. Presented at the annual meeting of the American Association of Clinical Immunologists and Allergists, New Orleans, LA. Mansfield, L. E. (1987). The role of food allergy in migraine: A review. Annals of Allergy, 58, 313-318. Mansfield, L. E., & Bowers, B. S. (1983). systemic reactions to papain in a nonoccupational setting. Journal of Allergy and Clinical Immunology, 71, 371­ 374. Mansfield, L. E., Vaughan, T. R., Waller, S. F., Haverly, R. W., & Ting, S. (1985). Food allergy and adult 259 migraine: Double-blind and mediator confirmation of an allergic etiology. Annals of Allergy, 55, 126­ 129. Maslansky, L., & Wein, G. (1971). Chocolate allergy: A double-blind study. Connecticut Medicine, 35, 5-9. Mason, J. W. (1975). Emotions as reflected in patterns of endocrine integration. In L. Levi (Ed.), Emotions: Their parameters and measurement. New York: Raven Press. Matje, D. (1984). stress and cancer: A review of the literature. Cancer Nursing, 2, 399-404. Mattes, J. A. (1983). The Feingold Diet: A current reappraisal. Journal of Learning Disabilities, 16, 319-323. Mattes, J.A., & Gittelman,' R. (1981). Effects of artificial food colorings in children with hyperactive symptoms. Archives of General Psychiatry, 38, 714-718. Mattes, J., & Gittelman-Klein, R. (1978). A crossover study of artificial food colorings in a hyperkinetic child. American Journal of psychiatry, 135, 987-988. Maulitz, R.M., Pratt, p.S., Shockett, A.L. (1979). Exercise-induced anaphylactic reaction to shellfish. Journal of Allergy and Clinical Immunology, 63, 433­ 434. 260 May, C. D. (1975). Food allergy. Pediatric Clinics of North America, 22, 217-220. May, C. D. (1976). Objective clinical and laboratory studies of immediate hypersensitivity reactions to foods in asthmatic children. Journal of Allergy and Clinical Immunology, 58, 500-515.

May, C. D. (1979). Food sensitivity. Paeditrician,~, 141-153. May, C. D. (1980). Food allergy - Material and ethereal. New England Journal of Medicine, 302, 1142-1143. May, C. D. (1982). Food allergy: Lessons from the past. Journal of Allergy and Clinical Immunology, 69, 255­ 259. May, C. D. (1984). Food sensitivity: Facts and fancies. Nutrition Reviews, 42, 1984. May, C. D. (1985). Are confusion and controversy about food hypersensitivity really necessary? Journal of Allergy and Clinical Immunology, 75, 329-333. May, C. D. (1986). Defined versus ill-defined syndromes associated with food sensitivity. Journal of Allergy and Clinical Immunology, 78, 144-148. May, C. D., & Bock, s. A. (1978). Adverse reactions to food due to hypersensitivity. In C. Reed and E. Ellis (Eds.), Allergy principles and practice. st. Louis: C. V. Mosby Company. 261 Mayron, L. W. (1979). Allergy, learning, and behavior problems. Journal of Learning Disabilities, 12, 41­ 51. McCarty, E. P., & Frick, O. L. (1983). Food sensitivity: Keys to diagnosis. The Journal of Pediatrics, 102, 645-652. McClelland, D. C., Alexander, C., & Marks, E. (1982). The need for power, stress, immune function, and illness among male prisioners. Journal of Abnormal Psychology, 91, 61-70. McClelland, D. C., Floor, E., Davidson, R. J., & Saron, C. (1980). stressed power motivation, sympathetic activation, immune function, and illness. Journal of

Human stress, ~, 11-19. McGrath, J. E. (Ed.). (1970). Social and psychological factors in stress. New York: Holt, Rinehart & Winston. McLoughlin, J.A., & NaIl, M. (1988). Teacher opinion of the role of food allergy on school behavior and aChievement. Annals of Allergy, 61, 89-91. McNeil, D., & Strauss, R.H. (1988). Exercise-induced anaphylaxis related to food intake. Annals of Allergy, 61, 440-442. Medleau, L., Latimer, K. S., & Duncan, J. R. (1986). Food hypersensitivity in a cat. Journal of the American Veterinary Medical Association, 189, 692-693. 262 Metcalfe, D. D. (1984). Food hypersensitivity. Journal of Allergy and Clinical Immunology, 73, 749-762. Meyer, R. J., & Haggerty, R. J. (1962). streptococcal infections in families. Pediatrics, 29, 539-549. Meyerowitz, S., Jacox, R. F., & Hess, D. W. (1968). Monozygotic twins discordant for rheumatoid arthritis: A genetic, clinical and psychological study of 8 sets. Arthritis and Rheumatology, 11, 1­ 21. Milich, R., & Pelham, W. E. (1986). Effects of sugar ingestion on the classroom and play group behavior of attention deficit disordered boys. Journal of Consulting and Clinical Psychology, 54, 714-718. Miller, J. B. (1977). A double-blind study of food extract injection therapy: A preliminary report. Annals of Allergy, 38, 185-191. Minford, A. M. B., Macdonald, A., & Littlewood, J. M. (1982). Food intolerance and food allergy in children: A review of 68 cases. Archives of Disease in Childhood, 57, 742-747. Moneret-Vautrin, D. A. (1986). Food antigens and additives. Journal of Allergy and Clinical Immunology, 78, 1039-1046. Monjan, A. A., & Collector, M. I. (1977). stress-induced modulation of the immune response. Science, 196, 307-308. 263 Monro, J., Brostoff, J., Carini, c., & Zilkha, K. (1980). Food allergy in migraine. Lancet, July 5, 1-4.

Moore, I. H. (1954). Infant food allergy. Lancet, 74, 80.

National Advisory Committee on Hyperkinesis and Food

Additives: Report to the Nutrition Foundation. New York, Nutrition Foundation, 1975. National Advisory Committee on Hyperkinesis and Food Additives (1980). Report to the Nutrition

Foundation, New York. Neuman, I., Elian, R., Nahum, H., Shaked, P., & Creter, D. (1978). The danger of 'yellow dyes' (tartrazine) to

allergic sUbjects. Clinical allergy, 8, 65-68. Novey, H.S., Fairshter, R.D., Salness, K., Simon, R.A., & Curd, J.G. (1983). Postprandial exercise-induced anaphylaxis. Journal of Allergy and Clinical

Immunology, 71, 498-504.

O'Banion, D. R., & Greenberg, M. R. (1982). Behavioral effects of food sensitivity. International Journal

of Biosocial Research, ~, 55-68. Ogle, K. A., & Bullock, J. D. (1977). Children with allergic rhinitis and/or bronchial asthma treated

with elimination diet. Annals of Allergy, 39, 8-11

Onorato, J., Merland, N., Terral, C., Michel, F. B., & Bousquet, J. (1986). Placebo-controlled double-blind

food challenge in asthma. Journal of Allergy and

Clinical Immunology, 78, 1139-1146. 264 O'Shea, J. A., & Porter, S. F. (1981). Double-blind study of children with hyperkinetic syndrome treated .with mUlti-allergen extract sUblingually. Journal of

Learning Disabilities, 14, 189-237. Palmer, S., Rapoport, J. L., & Quinn, P. O. (1975). Food additives and hyperactivity. Clinical Pediatrics,

14, 956-959. Panush, R. S., & Webster, E. M. (1985). Food allergies and other adverse reactions to foods. Medical Clinics of

North America, 69, 5~3-546. Parker, S.L., Sussman, G.L., & Krondl, M. (1988). Dietary aspects of adverse reactions to foods in adults. Canadian Medical Association Journal, 139, 711-718.

Peace, S. E., & Love, C. T. (1986). Optimal methods and issues in nutritional research in the correctional

setting. Nutrition Reviews, 44, 122-132.

Pearson, D. J. (1986). Pseudo food allergy. British

Medical Journal, 292, 221-222.

Pearson, D. J., & McKee, A. (1985). Food allergy.

Advances in Nutritional Research, ~, 1-37.

Pelikan, Z., & Pelikan-Filipek, M. (1987). Bronchial response to food ingestion challenge. Annals of

Allergy, 58, 164-172.

Peters, L. J. (1975). Enhancement of syngeneic murine

tumour transplantation by whole body irradiation - a 265 nonimmunological phenomena. British Journal of Cancer, 31, 293-300. Peters, L. J., & Kelly, H. (1977). The influence of stress hormones on the transplantability of a non­ immunogenic syngeneic murine tumour. Cancer, 39, 1482-1488. Pratt, E. L. (1958). Food allergy and food intolerance in relation to the development of good eating habits. Pediatrics, Apr., 642-648. Prinz, R. J., Roberts, W. A., & Hantman, E. (1980). Dietary correlates of hyperactive behavior in children. Journal of consulting and Clinical Psychology, 48, 760-769. Prinz, R. J., Roberts, W. A., & Hantman, E. (1981). Double-blind study implicates sugar, questions Feingold Diet. International Journal of Biosocial

Research, ~, 13-14. Randolph, T. G. (1976a). Historical development of

clinical ecology. In L. D. Dickey (Ed.), Clinical ecology (pp. 9-17). springfield, Illinois: Charles C. Thomas.

Randolph, T. G. (1976b). Adaptation to specific environmental exposures enhanced by individual susceptibility. In L. D. Dickey (Ed.), Clinical

ecology (pp. 46-66).' springfield, Illinois: Charles

C. Thomas. 266 Randolph, T.G. (1976c). Ecologically oriented history taking. In L.D. Dickey (Ed.), Clinical ecology (pp.

381-396). Springfield, Illinois: Charles C. Thomas.

Randolph, T. G. (1978). specific adaptation. Annals of

Allergy, 40, 333-345. Randolph, T. G., & Moss, R. W. (1980). An alternative approach to allergies. New York: Bantam Books. Rapp, D. J. (1978a). Does diet affect hyperactivity? Journal of Learning Disabilities, 11, 56-62.

Rapp, D. J. (1978b). Double-blind confirmation and

treatment of milk sensitivity. Medical Journal of

Australia, 1, 571-572. Rapp, D. J. (1979a). Allergies and the hyperactive child.

New York: Sovereign Books.

Rapp, D. J. (1979b). Food allergy treatment for hyperkinesis. Journal of Learning Disabilities, 12,

608-616.

Rapp, D. J. (1981). Clinical ecology: Recognition and

treatment of unsuspected food and environmental

allergies. International Journal of Biosocial

Research, 2, 29-36. Rea, W.J., & Ross, G.H. (1989). Foods and chemicals as environmental excitants. Nurse Practitioner, 14, 17- 40.

Reed, B. (1977). Diet related to killer diseases. U.s. Senate Select Committee on Nutrition and Human Needs, 267 95th Congress. Washington, D.C.: U.S. Government Printing Office. Reiriherz, E. L., & Schlossman, S. F. (1980). Current concepts in immunology: Regulation of the immune response - Inducer and suppressor T-lymphocyte subsets in human beings. New England Journal of Medicine, 303, 370-373. Reite, M., Harbeck, R., & Hoffman, A. (1981). Altered cellular immune response following peer separation. Life Sciences, 29, 1133-1136. Riley, V. (1975). Mouse mammary tumors: Alteration of incidence as apparent function of stress. Science, 189, 465-467. Riley, V. (1981). Psychoneuroendocrine influences on immunocompetence and neoplasia. Science, 212, 1100­ 1109.

Rimland, B. (1983). The Feingold Diet: An assessment of the reviews by Mattes, Kovale and Forness and others. Journal of Learning Disabilities, 16, 331-333. Rimland, B., & Larson, G. E. (1981). Nutritional and ecological approaches to the reduction of criminality, delinquency and violence. Journal of Applied Nutrition, 33, 118-128. Rinkel, H. J., Lee, C. Ho, Brown, Do Wo, Willoughby, J. W., & Williams, J. M. (1964). The diagnosis of food allergy. Archives of otolaryngology, 79, 71. 268 Rinkel, H. J., Randolph, T. G., & Zeller, M. (1951). Food allergy. Springfield, Illinois: Charles C. Thomas. Rippere, V. (1983). Food additives and hyperactive children: A critique of Conners. British Journal of Clinical Psychology, 22, 19-32. Rix, K. J. B., Pearson, D. J., & Bentley, S. J. (1984). A psychiatric study of patients with supposed food allergy. British Journal of Psychiatry, 145, 121­ 126. Rogers, M. P., Dubey, D., & Reich, P. (1979). The influence of the psyche and the brain on immunity and disease susceptibility: A critical review. Psychosomatic Medicine, 41, 147-164. Rose, T. L. (1978). The functional relationship between artificial food colors and hyperactivity. Journal of Applied Behavior Analysis, 11, 439-446. Rosen, L.A., Bender, M.E., Sorrell, S., Booth, S.R., McGrath, M.L., & Drabman, R.S. (1988). Effects of sugar (sucrose) on children's behavior. Journal of ConSUlting and Clinical Psychology, 56, 583-589. Rossiter, M. A. (1985). Food intolerance -A general pediatrician's view. Journal of the Royal Society of Medicine, 78, 17-20. Rowe, A. H. (1928). Food allergy: Its manifestations, diagnosis and treatment. Journal of the American Medical Association, 91, 1623-1630. 269 Rowe, A. H. (1930). Gastrointestinal food allergy. A study based on one hundred cases. Journal of Allergy, 1, 172-177.

Rowe, A. H. (1931). Food allergy (Its manifestations. diagnoses. and treatment). Philadelphia: Lea and Febiger. Rowe, A. H. (1937). Clinical allergy: Manifestations. diagnosis and treatment. Philadelphia: Lea & Febiger.

Rowe, A. H. (1950). Allergic toxemia and fatigue. Annals

of Allergy, ~, 72-84. Rowe, A. H., & Rowe, A. (1972). Food allergy: Its manifestations and control and the elimination diets. springfield, Illinois: Charles C. Thomas. Rowe, K.S. (1988). synthetic food coloruings and 'hyperactivity': A double-blind crossover study. Australian Paediatric Journal, 24, 143-147. Rubin, J. M., Shapiro, J., Muehlbauer, P., & Grolnick, M. (1965). Shock reaction following ingestion of mango. Journal of the American Medical Association, 193, 147-148.

Salzman, L. K. (1976). Allergy testing, psychological assessment and dietary treatment of the hyperactive

child syndrome. Medical Journal of Australia, ~,

248-251. 270 Sampson, H. A. (1983). Role of immediate food

hypersensitivity in the pathogenesis of atopic

dermatitis. Journal of Allergy and Clinical

Immunology, 71, 473-480.

Sampson, H. A., & Albergo, R. (1984). Comparison of results of skin tests, RAST, and double-blind,

placebo-controlled good challenges in children with

atopic dermatitis. Journal of Allergy and Clinical

Immunology, 74, 26.

Sampson, H. A., Buckley, R. H., & Metcalfe, D. D. (1987). Food allergy. Journal of the American Medical

Association, 258, 2886-2890.

Schauss, A. G. (1980). Diet, crime and delinquency.

Berkeley: Parker House. Schauss, A. G., & Simonsen, C. E. (1979). Critical

analysis of the diets of ~hronic juvenile offenders,

Part 1. Journal of Orthomolecular Psychiatry, ~,

149-157.

Schleifer, S. J., Keller, S. E., Camerinio, M., Thornton, J. C., & stein, M. (1983). Suppression of lymphocyte stimulation following bereavement. Journal of the

American Medical Association, 250, 374-377.

Schleifer, S. J., Scott, B., Stein, M., & Keller, S. E. (1986). Behavioral and developmental aspects of

immunity. Journal of the American Academy of Child

Psychiatry, 25, 751-763. 271 Schloss, o. M. (1912). A case of allergy to common foods.

American Journal of Diseases of Children, ~, 341-362. Schneider, W. F. (1945). Psychiatric evaluation of the hyperkinetic child. Journal of Pediatrics, 26, 559­ 570. Schoenthaler, S. J. (1982). The effect of sugar on the treatment and control of antisocial behavior: A double-blind study of an incarcerated juvenile population. International Journal of Biosocial

Research, ~, 1-9. Schoenthaler, S. J. (1983a). Diet and delinquency: A multistate replication. International Journal of Biosocial Research, a, 70-78. Schoenthaler, S. J. (1983b). The Alabama diet-behavior program: An empirical evaluation at the Coosa Valley Regional Detention Center. International Journal of Biosocial Research, a, 79-87. Schoenthaler, S. J. (1983c). The Los Angeles Probation Department diet-behavior program: An empirical analysis of six institutional settings. International Journal of Biosocial Research, 5, 88­ 98. Schoenthaler, S. J. (1983d). The Northern California diet­ behavior program: An empirical examination of 3,000 incarcerated juveniles in stanislaus County Juvenile 272 Hall. International Journal of Biosocial Research, 2, 99-106.

Schoenthaler, S. J. (1983e). The effects of citrus on the

treatment and control of antisocial behavior: A

double-blind study of an incarcerated juvenile

population. International Journal of Biosocial Research, 2, 107-117. Schoenthaler, S. J. (1985). Diet and delinquency:

Empirical testing of seven theories. International

Journal of Biosocial Research, 1, 108-131.

Schoenthaler, S. J. Nutrition and antisocial behavior.

Unpublished manuscript.

Schofield, A. T. (1908). A case of egg poisoning. Lancet,

Mar. 7, 716. schreiber, R. A., & Walker, W. A. (1989). Food allergy: Facts and Fiction. Mayo Clinic proceedings, 64,

1381-1391.

Seligman, M. E. (1975). Helplessness: On depression,

development, and death. San Francisco: Freeman.

Sell, S. (Ed.). (1987). Immunology, immunopathology, and

immunity. New York: Elsevier.

Selye, H. (1936). A syndrome produced by diverse nocuous

agents. Nature, 138, 32.

Selye, H. (1946). The general adaptation syndrome and the

diseases of adaptation. Journal of Clinical

Endocrinology and Metabolism, ~, 117-130. 273 Selye, H. (1955). stress and disease. Science, 122, 625- 631. Selye, H. (1976). The stress of life. New York: McGraw­ Hill Book Co. settipane, G. A. (1986). The restaurant syndromes. Archives of Internal Medicine, 146, 2129-2130. Shannon, W. R. (1922). Neuropathic manifestations in infants and children as a result of anaphylactic reaction to foods contained in their diet. American Journal of Diseases of Children, 24, 89-94. Shaywitz, B A., Goldenring, J. R., & Wool, R. S. (1978). The effects'of chronic administration of food colorings on activity levels and cognitive performance in normal and hyperactive developing rat

pups. Annals of Neurology, ~, 196. Sheridan, J. J., & Meister, K. A. (1982). Food additives and hyperactivity. New York: American Council on Science and Health. Sieben, R. L. (1977). Controversial medical treatments of learning disabilities. Academic Therapy, 13, 133­ 147. singh, M. M., & Kay, S. R. (1976). Wheat gluten as a pathogenic factor in schizophrenia. Science, 191, 401-402. 274 Sklar, L. S., & Anisman, H". (1979). Stress and coping factors influcence tumor growth. Science, 205, 513­ 515. Sklar, L. S., & Anisman, H. (1981). Stress and cancer. Psychological Bulletin, 89, 369-406. Sloan, A. E., & Powers, M. E. (1986). A perspective on popular perceptions of adverse reactions to foods. Journal of Allergy and Clinical Immunology, 78, 127­ 133. snyder, S. H., & Meyerhoff, J. L. (1973). How amphetamine acts in minimal brain dysfunction. Annals of the New York Academy of Sciences, 205, 310-320. Speer, F. (1954a). The allergic tension-fatigue syndrome.

Pediatric Clinics of North America, ~, 1029-1037. Speer, F. (1954b). Allergic tension-fatigue in children. Annals of Allergy, 12, 168-171. Speer, F. (Ed.). (1963). The Allergic Child. New York: Harper and Row. Speer, F. (1975). MUltiple food allergy. Annals of Allergy, 34, 71-76. Spielberger, C. D., & Sarason, I. G. (Eds.). (1977). Stress and anxiety. New York: Wiley.

Spring, C., Vermeersch, J., Blunden, D., & Sterling, H. (1981). Case studies of effects of artificial food colors on hyperactivity. Journal of Special Education, 15, 361-372. 275 stare, F. J., Whelan, E. M., & Sheridan, M. (1980). Diet and hyperactivity: Is there a relationship? Pediatrics, 66, 521-525. Staudenmayer, H., & Selner, J. C. (1987). Post-traumatic stress syndrome (PTSS): Escape in the environment. Journal of Clinical Psychology, 43, 156-157. stein, M., Keller, S. E., & Schleifer, S. J. (1985). Stress and immunomodulation: The roles of depression and neuroendocrine function. Journal of Immunology, 135, 827s-833s. Stein, M., & Schleifer, S. J. (1985). Frontiers of stress research: stress and immunity. In M. R. Zales (Ed.), Stress in health and disease (pp. 97-114). New York: Brunner/Mazel. Stine, J. J. (1976). Symptom alleviation in the

hyperactive child by diet~ry modification: A report on two cases. American Journal of Orthopsychiatry, 46, 637-645. Strickland, B. R. (1982). Implications of food and chemical susceptibilities for clinical psychology. International Journal of Biosocial Research, d' 39­ 43. Swanson, J. M., & Kinsbourne, M. (1980). Food dyes impair performance of hyperactive children on a laboratory learning test. Science, 207, 1485-1486. 276 Swenson, R. M., & Vogel, w. F. (1983). Plasma catecholamine and corticosterone as well as brain catecholamine changes during coping in rats exposed to stressful footschock. Pharmacology. Biochemistry & Behavior, 18, 689-693. Tache, Y., DuRuisseau, P., Tache, J., Selye, H, & Collu, R. (1976). Shift in adenohypophyseal activity during chronic intermittent immobilization of rats. Neuroendocrinology, 22, 325-336. Taylor, E. (1979). Food additives, allergy and hyperkinesis. Journal of Child Psychology and Psychiatry, 20, 357-363. Taylor, E. (1984). Diet and behavior. Archives of Disease in Childhood, 59, 97-98. Taylor, S. L. (1986). Food allergies and sensitivities. Journal of the American Dietetic Association, 86, 599-600. Taylor, S. L., Bush, R. K., & Busse, W. W. (1986). Avoidance diets - How selective should we be? New England and Regional Allergy proceedings, Z, 527-532. Terr, A. I. (1986). Environmental illness. Archives of Internal Medicine, 146, 145-149. Teshima, H., Kubo, C., Kihara, H., Imada, Y., Nagata, S., Ago, Y., & Ikemi, Y. (1982). Psychosomatic aspects of skin diseases from the standpoint of immunology. Psychotherapy and Psychosomatics, 37, 165-175. 277 Thorley, G. (1983). Childhood hyperactivity and food additives. Developmental Medicine and Child Neurology, 25, 531-534.

Trumbull, R., & Appley, M. H. (1986). A conceptual model for the examination of stress dynamics. In M. H. Appley & R. Trumbull (Eds.), Dynamics of stress (pp. 21-45). New York: Plenum Press. Truswell, A. S. (1985). Food sensitivity. British Medical Journal, 291, 951-955. Tryphonas, H. (1979). Factors possibly implicated in hyperactivity: Feingold's hypothesis and hypersensitivity reactions. In R. Trites (Ed.), Hyperactivity in Children: Etiology, Measurement, and Treatment Implications. Baltimore: University Park Press. Tryphonas, H., & Trites, R. (1979). Food allergy in children with hyperactivity, learning disabilities and/or minimal brain dysfunction. Annals of Allergy,

42, 22-27.

Udelman, H. 0., & Udelman, D. L. (1983). Current explorations in psychoimmunology. American Journal of Psychotherapy, 37, 210-221.

Vane, J. R. (1971). I~hibition of prostaglandin synthesis as a mechanism of action for aspirin-like drugs. Nature (New Biology), 231,232-235. 278 Varley, C. K. (1984). Diet and the behavior of children with attention deficit disorder. Journal of the American Academy of Child Psychiatry, 23, 182-185. Veno, A., & Davidson, M. J. (1978). A relational model of

stress and adaptation. Man Environment Systems, ~, 75-89. Vercruyssen, M. (1984). Breathing 3-4% C02: Behavioral effects. In E. Kamon, N. S. Deno, and C. F. Ryan (Eds.), Physiological responses of miners to emergency. End of contract report for Bureau of Mines Contract No. JO-100092. Washington, D. C.: U. S. Department of the Interior. Vercruyssen, M. (1989). Selective effects of carbon dioxide information processing: Additive factors method in behavioral·toxicoloy. Manuscript submitted for pUblication. Walker, S. (1977). Help for your hyperactive child. Boston: Houghton Mifflin. Wall, H. C. (1987). Letter to the editor. Annals of Allergy, 58, 149-150. Warner, J., & Hathaway, M. J. (1984). Allergic form of Meadow's syndrome (Munchausen by proxy). Archives of Disease in Childhood, 59, 151-156. Weiner, H. (1977). Psychobiology and human disease. New York: Elsevier. 279 Weiss, B. (1982). Food additives and environmental chemicals as sources of childhood behavior disorders. Journal of the American Academy of Child Psychiatry,

21, 144-152.

Weiss, B. (1986). Food additives as a source of behavioral

disturbances in children. Neurotoxicology,~, 197­

208. Weiss, B., Williams, J. H., Margen, S., Abrams, B., Caan, B., citron, L. J., Cox; C., McKibben, J., agar, D., & Schultz, S. (1980). Behavioral responses to artificial food colors. Science, 207, 1487-1489. Weiss, G., Minde, K., Werry, J. S., Douglas, V., & Nemeth, E. (1971). Studies on the hyperactive child. Archives of General psychiatry, 24, 409-414.

Wender, E. H. (1977). Food additives and hyperkinesis. American Journal of Diseases in Children, 131, 1204­

1206. Wender, E. (1980). New evidence on food additives and hyperkinesis. American Journal of Diseases of

Children, ld!, 1122-1128. Wender, E. H. (1986). The food additive-free diet in the treatment of behavior disorders: A review. Developmental and Behavioral Pediatrics, 1, 35-42. Wender, P. H. (1973). Some speculations concerning a possible biochemical basis of minimal brain 280 dysfunction. Annals of the New York Academy of Sciences, 205, 18-28. Werry, J. S. (1976). Food additives and hyperactivity.

Medical Journal of Australia, ~, 281-282. Werry, J. S., Sprague, R. L., & Cohen, M. N. (1975). Conner's teacher rating scale for use in drug studies with children - An empirical study. Journal of Abnormal Child Psychology, d' 217-229. Wilken-Jensen, K., & Melchior, J. C. (1970). Food allergy, anaphylactic shock and neurological sYmptoms. Annals of Allergy, 28, 539-542. Williams, J. I., Cram, D. M., Tausig, F. T., & Webster, E. (1978). Relative effects of drugs and diet on

hyperactive behaviors: An experimental study. Pediatrics, 61, 811-817. Wilson, N., & Scott, A. (1989). A double-blind assessment of additive intolerance in children using a 12 day challenge period at home. Clinical and Experimental

Allergy, ~, 267-272. Winkelman, N. W., & Moore, M. T. (1941). Allergy and nervous diseases. Journal of Nervous and Mental

Disease, 93, 736-749~ Wolraich, M., Milich, R., Stumbo, P., & Schultz, F. (1985). Effects of sucrose ingestion on the behavior of hyperactive boys. Journal of Pediatrics, 106, 675­ 682. 281 Wolraich, M. L., stumbo, P. J., Milich, R~, & Schultz, F. (1986). Dietary characteristics of hyperactive and control boys. Journal of the American Dietetic Association, 86, 500-504. Workman, E. A., & La Via, M. F. (1987). T-lymphocyte polyconal proliferation: Effects of stress and stress response style on medical students taking National Board Examinations. Clinical Immunology and Immunopathology, 43, 308-313. Yen, P. K. (1987). Food allergy facts. Geriatric Nursing, Jan/Feb, 35.

Young, E., Patel, S., Stoneham, M., Rona, R., & Wilkinson, J. D. (1987). The prevalence of reaction to food additives in a survey population. Journal of the Royal College of Physicians of London, 21, 241-247. Yuwiler, A. (1976). stress, anxiety and endocrine function. In R. G. Greenell & s. Gabay (Eds.), Biological foundations of psychiatry. New York: Raven Press.