Impact of Food Immunoglobulin G-Based Elimination Diet on Subsequent Food Immunoglobulin G and Quality of Life in Overweight/Obese Adults

Abstract

Objectives:

The goal of this study was to assess changes in serum immunoglobulin G (IgG) food antibody titers and quality-of-life measurements following a targeted elimination diet in overweight/obese adults.

Methods:

We performed a randomized control trial. Participants were randomized in a 2:1 ratio to either an intervention group or waitlist group for 3 months. Food IgG testing was performed on all participants. The intervention group was instructed to eliminate up to 10 foods, for which they had high titers of IgG and communicated with health coaches for nutritional counseling for meal planning and adherence. The waitlist group did not receive their IgG testing results or health coaching. Primary outcome was serum IgG titers for foods eliminated during the trial, compared with baseline concentrations. Secondary outcomes were health-related quality of life measured by Patient-Reported Outcomes Measurement Information System (PROMIS-29) and change in participant-identified symptom severity measured by Measure Yourself Medical Outcome Profile. Exploratory outcomes were changes in body weight and waist circumference.

Results:

IgG antibody concentrations decreased in 83% of the targeted foods in the treatment group and in 60% of the foods in the waitlist group, but this was not found to be a statistically significant difference. The intervention group reported improvement in sleep during the trial compared with waitlist, which was the only statistically significant finding in the study.

Conclusions:

The findings are consistent with changes in IgG titer measurements following an elimination diet based on IgG testing. Future larger clinical trials are necessary to determine the degree to which these findings are generalizable.

Introduction

Adults with food allergies or sensitivities report a wide range of symptoms, including gastrointestinal distress, migraines, arthritis, rashes, and respiratory distress in addition to a reduced quality of life due to these symptoms.^1,2 Approximately 15%–20% of adults report a food sensitivity or allergy, with increasing frequency over the past 15 years.³ Of these adults, only about 1/10th have an immunoglobulin E (IgE)-mediated allergic response.⁴ IgE-mediated responses are characteristic of atopic allergic reactions and include classic symptoms, such as hives, difficulty breathing, abdominal pain, and itchiness in and around the mouth. Because of the small percentage of IgE-mediated allergies, IgE testing has not proven useful for detecting the majority of food sensitivities.⁵

Immunoglobulin G (IgG) antibodies, on the other hand, have been hypothesized as a contributor to the symptoms associated with food sensitivities but research remains limited and inconclusive.⁶ In addition to clinical observations suggesting a role in food sensitivities, it has been hypothesized that high concentrations of IgG antibodies result from increased intestinal epithelial permeability to larger proteins, triggering an immunological reaction.^6,7

The clinical gold standard for confirming food sensitivity is a strict elimination diet followed by a challenge diet involving a slow, one-by-one systematic reintroduction of individual foods.⁸ This process can be difficult and time consuming for patients, and adherence compliance with these diets is poor. Furthermore, food sensitivities may be difficult for patients to self-identify as it may take hours or days after ingestion to trigger a symptomatic reaction.⁵ Thus, IgG antibody testing may be a reasonable tool to help target the elimination diet process and/or aid health care providers and patients in identifying food sensitivity reactions.

The prevalence of obesity in the United States has more than doubled since the 1960s; currently, over two-thirds of adults are considered overweight or obese.⁹ Several factors affect this growing health crisis, including individual genetics, food availability, dietary choices, stress, physical activity, income, and mental health.⁹ Overweight and obese adults are at greater risk for certain health problems, such as diabetes, hypertension, heart disease, fatty liver disease, osteoarthritis, mental health disorders, and some cancers.^10,11 Increased inflammation has been theorized as a mediator of many of the health effects of obesity, including hypertension, cardiovascular disease, and diabetes.¹² Inflammation is also considered an important causative factor in the dysregulation of the mucosal barrier in the gastrointestinal tract.^7,13 The resulting conceptual framework includes inflammation as a cause and effect of overweight/obesity, in turn contributing to increased intestinal epithelial dysfunction, potentially increasing measurable IgG antibodies to food.^12,14,15 This conceptual model is supported by the work of Lewis et al. who conducted an IgG antibody-based elimination diet for 120 participants with a body–mass index (BMI) >20 kg/m² demonstrating reduced BMI and waist circumference.¹⁶

Behavioral support for adjusting nutritional practices has been shown to be effective in helping participants adhere to dietary recommendations.^17,18 In a study examining behavioral support with weight loss in 415 obese participants, those who received either in-person or telephone support were significantly more likely to lose weight.¹⁸ Those receiving remote support through telephone achieved similar weight loss results to those who received in-person support.

To inform research gaps related to the effects of an IgG testing-targeted elimination diet on IgG titers, quality of life and subsequent weight loss, we conducted a randomized trial of IgG-targeted elimination diets in overweight or obese adults.

Materials and Methods

Design and participants

This randomized control trial was approved by the Institutional Review Board at the National University of Natural Medicine. Both men and women were recruited from the Portland, OR metropolitan area. Eligible participants included those 18–65 years of age with a BMI of ≥25 kg/m². Each participant had to be willing to follow an elimination diet based on their IgG test results for 3 months. The trial was limited to participants who spoke and read English and to those who had access to the internet to complete online surveys. Participants were excluded if they were pregnant, on any immunosuppressive therapies in the last 2 weeks, or on current long-term antibiotic use (>21 days). Participants who did not have any IgG food sensitivities (defined as those in class 2 or 3 categories) were excluded from the trial.

Eligible candidates (n = 30) were randomized in a 2:1 ratio of intervention to the waitlist group. Adherence in diet studies are typically low; therefore, we increased our target sample size in the intervention group to adjust for expected high rate of attrition.¹⁹ Assuming a 20% rate of reduction in sensitivity to targeted foods in the control group, we were powered to detect an absolute increase of 20% (to 40% or more) in the treatment group, relative to control.

Measures

IgG food testing

IgG antibodies were measured using ELISA methodology (Alletess Medical Laboratory), which detects all sub-fractions of IgG (i.e., IgG1, IgG2, IgG3, and IgG4).²⁰ Food antigens were mixed with participant's serum and IgG antibody complexes were then measured on a continuous scale. The amount of IgG was divided into four categories (class 0, 1, 2, 3). The higher the concentration of IgG detected, the higher the class. For the purpose of this study a positive IgG result was considered to be present for all IgG scores in class 2 or 3 at baseline.

A food frequency questionnaire (FFQ) was administered online to participants at baseline, 1-, 2-, and 3-month follow-up to determine which food exposures the participant had encountered before and during the trial. The FFQ was designed based on the question structure from the National Health and Nutrition Examination Survey customized for the foods included in the IgG test, and anchored to the last 90 days for the baseline questionnaire, and the last 30 days for the follow-up.²¹

Questionnaires

The Patient-Reported Outcomes Measurement Information System (PROMIS-29) Profile instrument was administered at baseline and at 3-month follow-up. The PROMIS-29 Profile is a validated health-related quality-of-life questionnaire, including seven different subdomains (Fatigue, Depression, Anxiety, Pain Interference, Sleep Disturbance, Physical Function, and Ability to Participate in Social Roles and Activities). Each subdomain has four different questions rated on a 5-point scale for a total of 20 points possible. An additional question asks the user to rate their pain intensity on a scale from 0 to 10.

Measure Yourself Medical Outcome Profile (MYMOP), a patient-centered outcome measure that allows for the participant to self-select their top health concern, was administered at baseline and 3-month follow-up.^22,23 Each participant was asked to use their own words to select the top symptom they are concerned about and optionally to choose a related second symptom. The participant could also choose an activity limited by the health concern. Each of these questions are scored on a 0–6 scale, with higher scores representing more severe symptoms. Scoring of the MYMOP is an average of all of the profile questions answered.

Body weight and waist circumference were exploratory outcomes for this study.

Intervention

IgG antibody titers were measured for 96 different foods. Participants in the intervention group were given their food sensitivity data at the beginning of the trial and asked to eliminate up to 10 foods. Food sensitivities were defined as those that had the highest concentration of IgG reaction (class 2 or 3). Up to 10 foods that fell in class 2 or 3 were identified as the targeted foods for elimination during the trial. If a participant had more than 10 class 2 or 3 foods, they were asked to eliminate the 10 foods with the highest IgG concentration. Intervention group participants were offered telephone nutritional counseling sessions with registered dieticians at the beginning of the trial, and up to 10 additional sessions throughout the trial. Up to 10 targeted foods (class 2 or 3) were also identified in the waitlist group for comparison to the treatment group. Waitlist participants did not receive their results until 3 months after their enrollment. At that time, they were retested, given both sets of data and offered an opportunity to discuss their results with a registered dietician.

Data analyses

Changes in IgG titers within an individual participant were computed both continuously, as the mean of changes in raw measures across targeted foods; and as either the number or percentage of targeted foods showing reductions in sensitivity class.

Mean intraindividual change in IgG was compared between groups using an independent two-sided, t-test. We also compared changes for all individual foods between groups using a mixed effects model with participant as a random effect. Numbers of targeted foods, numbers of targeted foods showing categorical class reductions, and percentage of targeted foods showing class reductions were all compared between groups, using independent t-tests or nonparametric Mann–Whitney tests, as appropriate.

In secondary analyses, we compared mean within-individual changes in sensitivity between participants in the intervention group who either (1) did or did not completely adhere to the dietary recommendations; or (2) mostly adhered versus did not adhere. The FFQ were used to determine adherence to the elimination diet. The participants who completely eliminated each of their targeted foods were considered completely adherent to the intervention. Mostly adhering was defined as reducing (but not necessarily eliminating) intake of all targeted foods. Given the even smaller sample sizes for these comparisons, they were made using independent t-tests.

Additionally, changes in Pain Intensity and MYMOP composite scores were compared between groups using Mann–Whitney tests. Changes in PROMIS-29 domain scores and participant weight and waist circumference were compared using independent t-tests.

Analysis for the study was done first looking at only those participants that completed the trial. In addition, an intention-to-treat analysis was done. Noncompleters pre–post changes were imputed for all outcomes by sampling with replacement from the set of change scores available from the waitlist group. Imputation was carried out 10 times and mean and standard deviation for each group were summarized across the multiple imputations.²⁴

Results

Figure 1 summarizes enrollment in the clinical trial and Table 1 details baseline characteristics of participants by groups. As shown in Table 1, there was no significant difference between intervention and waitlist groups with regard to age, sex, race, or educational background.

FIG. 1.

Summary of enrollment.

Table 1.

Baseline Characteristics of Treatment and Waitlist Groups

	All participants at baseline			Participants who completed trial
Characteristics	Treatment (n = 20)	Waitlist (n = 10)	p-Value for difference	Treatment (n = 15)	Waitlist (n = 9)	p-Value for difference
Age (years)	41.4 ± 12.9	41.2 ± 14.1	>0.50	42.2 ± 12.7	41.0 ± 14.9	>0.50
Sex, n (%)
Female	18 (90)	9 (90)	>0.50	14 (93.3)	8 (88.9)	>0.50
Male	2 (10)	1 (10)	>0.50	1 (6.7)	1 (11.1)	>0.50
Race, n (%)
White/Caucasian/European	15 (75)	7 (70)	>0.50	12 (80.0)	6 (66.7)	>0.50
Black/African American	1 (5)	1 (10)		1 (6.7)	1 (11.1)
Asian/Asian American	1 (5)	1 (10)		0	1 (11.1)
More than one	2 (10)	1 (10)		2 (13.3)	1 (11.1)
Unknown/not reported	1 (5)	0
Education level, n (%)
Some college	3 (15)	2 (20)	>0.50	1 (6.7)	2 (22.2)	0.45
2-Year college	2 (10)	2 (20)		1 (6.7)	1 (11.1)
4-Year college	7 (35)	3 (30)		5 (33.3)	3 (33.3)
Some graduate school	4 (20)	0		4 (26.7)	0
Graduate degree	4 (20)	3 (30)		4 (26.7)	3 (33.3)

A summary of the 10 foods with greatest frequency of measurable IgG antibodies is shown in Figure 2. Ten participants (42%) were found to have detectable IgG to egg (yolks or whites). Cow's milk was also a common sensitivity with 10 participants (38%). Safflower and sesame were equally common with six (25%) participants each. The average number of IgG-positive foods in each group was similar between treatment (4.8 ± 3.3) and waitlist (4.1 ± 3.3).

FIG. 2.

Frequencies of IgG-positive foods at baseline. *Includes egg yolks and/or egg whites. **Includes cheddar cheese, cottage cheese, cow's milk, mozzarella cheese, and/or yogurt. IgG, immunoglobulin G.

Regarding use of nutritional coaches over the course of the study, all participants spoke with the counselors only during the first 3 weeks, and participants averaged only 1.9 ± 1.0 such visits. Less than half (7/15) of the participants in the group completely eliminated the recommended foods for the entire 3 months.

Using data from only participants with follow-up measures (Table 2), there was a higher mean number of foods in the intervention group showing a reduction in class (3.7 ± 2.6) versus waitlist (2.7 ± 2.4); however, this difference was not significant (p = 0.3). The median percentage of foods with a class reduction was higher in the intervention group (85.7% [66.7, 100], minimum 50%) compared with waitlist group (60% [25, 100], minimum 0%), but this comparison was likewise nonsignificant (p = 0.08 by independent t-test). When missing data was imputed to include data from all 30 randomized participants, the comparison between groups remained insignificant (Table 3). Within the treatment group, the reduction between participants who adhered to the diet versus those who only partially adhered were virtually the same (Table 2). Based on the FFQ, the treatment group participants reduced (but not necessarily eliminated) consumption of 92% of the targeted foods compared with 28% in the waitlist group. Just over 50% of the waitlist participants had reduced at least some of their targeted foods.

Table 2.

Comparison of Outcome Measures Between Groups for Completed Trial Participants

	Treatment (n = 15)	Waitlist (n = 9)	p
Reduction in IgG class (no. foods)	3.7 ± 2.6	2.7 ± 2.4	0.30
Reduction in IgG class (% foods)	85.7 [66.7, 100] minimum: 50	60 [25, 100] minimum: 0	0.08
Change in IgG scores	−0.091 ± 0.042	−0.056 ± 0.090	0.20
100% adhered to diet (n = 7)	−0.092 ± 0.031	—	—
100% adhered plus mostly adhered (n = 10)	−0.093 ± 0.036	—	—
Partially adhered to diet (n = 5)	−0.089 ± 0.036	—	—
MYMOP score change	−1.2 ± 1.25	−1.5 ± 0.87	>0.50
Weight change (lb)	−0.49 ± 9.14	−2.72 ± 5.65	>0.50
Waist change (inches)	−0.92 ± 3.08	−1.70 ± 2.81	>0.50

For reduction in IgG class (% foods) values are presented as median [IQR].

IgG, immunoglobulin G; MYMOP, Measure Yourself Medical Outcome Profile.

Table 3.

Comparison of Outcome Measures Between Groups for All Participants Imputed

	Treatment (n = 20)	Waitlist (n = 10)	p
Reduction in IgG class (no. foods)	3.3 ± 2.6	2.8 ± 2.4	>0.5
Reduction in IgG class (% foods)	78 [72, 82]	62.1 [59, 65]	0.19
Change in IgG scores	−0.086 ± 0.047	−0.059 ± 0.086	0.28
100% adhered to diet (n = 7)	—	—	—
100% adhered plus mostly adhered (n = 10)	—	—	—
Partially adhered to diet (n = 5)	—	—	—
MYMOP score change	−1.3 ± 1.17	−1.5 ± 0.86	>0.50
Weight change (lb)	−1.29 ± 8.48	−2.93 ± 5.66	>0.50
Waist change (inches)	−1.15 ± 3.04	−1.77 ± 2.69	>0.50

For reduction in IgG class (% foods) values are presented as mean (range).

IgG, immunoglobulin G; MYMOP, Measure Yourself Medical Outcome Profile.

Both treatment and waitlist groups showed a mean improvement in their individually identified symptoms as measured by MYMOP over the 3 months. There was no significant difference between the two groups (p > 0.3).

For the participants that completed the trial, the difference in sleep disturbance between groups was the only significant finding (p = 0.03) among PROMIS outcomes, with the intervention group demonstrating improved sleep quality compared with the waitlist. Imputing the missing data from all participants, sleep quality did not remain significant (p = 0.07). Comparisons of all other domains were nonsignificant (Tables 4 and 5).

Table 4.

Comparison of Patient-Reported Outcomes Measurement Information System Domains Between Treatment Groups for Completed Trial Participants

	Baseline	3 Months	Mean change	p^*
Physical function
Treatment (n = 15)	4.43 ± 0.75	4.52 ± 0.68	0.09 ± 0.42	0.50
Waitlist (n = 9)	4.42 ± 0.77	4.25 ± 0.98	−0.17 ± 1.21	0.50
Anxiety
Treatment (n = 15)	2.04 ± 0.58	2.05 ± 0.77	0.01 ± 0.67	>0.50
Waitlist (n = 9)	2.08 ± 1.07	1.89 ± 0.88	−0.17 ± 0.78	>0.50
Depression
Treatment (n = 15)	1.76 ± 0.76	1.77 ± 0.87	0.01 ± 0.66	>0.50
Waitlist (n = 9)	1.56 ± 0.82	1.53 ± 0.65	−0.03 ± 0.26	>0.50
Fatigue
Treatment (n = 15)	2.85 ± 1.08	2.32 ± 0.90	−0.53 ± 0.88	0.10
Waitlist (n = 9)	2.41 ± 1.25	2.56 ± 1.09	0.15 ± 1.01	0.10
Sleep disturbance
Treatment (n = 15)	2.58 ± 0.61	2.42 ± 0.88	−0.16 ± 0.76	0.03
Waitlist (n = 9)	2.35 ± 0.79	2.89 ± 0.92	0.54 ± 0.61	0.03
Ability to participate in social activities
Treatment (n = 15)	3.57 ± 1.16	3.89 ± 0.94	0.32 ± 0.69	0.22
Waitlist (n = 9)	3.92 ± 0.82	3.75 ± 1.37	−0.17 ± 1.23	0.22
Pain interference
Treatment (n = 15)	2.00 ± 1.15	1.42 ± 0.57	−0.58 ± 0.97	>0.50
Waitlist (n = 9)	2.11 ± 1.01	2.19 ± 1.18	0.08 ± 1.8	>0.50

p-Value for comparison of mean change between groups, by independent t-test.

Table 5.

Comparison of Patient-Reported Outcomes Measurement Information System Domains Between Treatment Groups for All Participants Imputed

	Baseline	Mean change	p ^*
Physical function
Treatment (n = 15)	4.54 ± 0.68	−0.05 ± 0.83	>0.50
Waitlist (n = 9)	4.47 ± 0.75	−0.13 ± 1.22	>0.50
Anxiety
Treatment (n = 15)	2.04 ± 0.72	0.07 ± 0.69	>0.50
Waitlist (n = 9)	1.95 ± 1.06	−0.14 ± 0.78	>0.50
Depression
Treatment (n = 15)	1.84 ± 1.03	0.01 ± 0.59	>0.50
Waitlist (n = 9)	1.5 ± 0.79	−0.03 ± 0.25	>0.50
Fatigue
Treatment (n = 15)	2.85 ± 1.06	−0.36 ± 0.95	0.21
Waitlist (n = 9)	2.27 ± 1.26	0.12 ± 0.98	0.21
Sleep disturbance
Treatment (n = 15)	2.66 ± 0.69	−0.03 ± 0.80	0.07
Waitlist (n = 9)	2.27 ± 0.79	0.57 ± 0.61	0.07
Ability to participate in social activities
Treatment (n = 15)	3.51 ± 1.07	0.21 ± 0.88	0.36
Waitlist (n = 9)	4.03 ± 0.85	−0.15 ± 1.23	0.36
Pain interference
Treatment (n = 15)	2.05 ± 1.17	−0.49 ± 0.9	0.13
Waitlist (n = 9)	2.00 ± 1.01	0.10 ± 1.2	0.13

p-Value for comparison of mean change between groups, by independent t-test.

Both groups showed a mean decrease in weight and waist circumference; however, there was tremendous variation, and the mean reductions were greater in the waitlist group (p > 0.5).

Discussion

IgG antibody testing has been utilized in some clinical practices as a way to identify suspected food sensitivities and to develop diets targeted to reduce symptoms in patients. Our study aimed to determine if an IgG-specific diet in an overweight population would lead to reduced IgG antibodies and/or improvements across a constellation of symptoms. Few studies have measured the IgG antibody concentrations over time following an elimination diet and most studies focus only on symptom changes. While the results did not demonstrate significant differences between the groups, a trend was measured that suggests eliminating or reducing a food leads to a decrease in the IgG antibody concentration for that individual food, which, although sounds obvious, has rarely been confirmed in formal research. Based on the results, complete elimination of the food may not be necessary to see the reduction in total IgG concentrations. In addition to an overall reduction in IgG concentrations in the intervention group, we found improved self-identified symptoms; however, these findings were not significantly different than the waitlist group. Despite the waitlist group not receiving their IgG panel results until after their follow-up visit, both groups exhibited class reductions in IgG-positive foods, MYMOP results, and biometric measures indicating the suggestion to make behavioral changes related to dietary choices could have inspired participants to make actual changes in behavior, subsequently impacting their overall health during the course of the study.

Exploratory measures in this study included monitoring changes in weight, BMI, and waist circumference. These measures trended down in both intervention and waitlist groups, but there was tremendous variation among participants. This study was not advertised as weight loss study and only a minority of participants chose to track their weight on their MYMOP, suggesting weight loss was not a key motivating factor for seeking to participate in this study. Additionally, nutritional counseling provided during the course of this study focused on eliminating IgG-positive foods instead of making nutritional choices for weight loss. Finally, because wide variety of factors influence weight loss, including both food choices and environmental influences, it is not surprising these measures did not differ between the two groups.²⁵ Even so, it appears that IgG-targeted diet modification did not directly facilitate weight loss.

Behaviors surrounding dietary choices are difficult to change.²⁶ Although behavioral support for adjusting nutritional choices has been shown to improve adherence to these changes, participants in the intervention group did not take advantage of the 10 sessions of nutritional counseling offered during the study.^17,18 Numerous studies have examined the psychological aspects of weight loss-related dieting, including motivation for change, predictors for success, and attitudes toward behavior change.²⁶ In this study, we found that many participants in the intervention group did not adhere completely to the elimination diet, and some participants actually increased their frequency of intake of IgG-positive foods. This is consistent with research suggesting that attempts to eat healthier may have the opposite effect.²⁷ In one study, restrained eaters exhibited increased craving or desire for particular foods and increased consumption of those foods as compared with unrestrained eaters. Therefore, it is plausible some participants in the intervention group were unsuccessful in eliminating IgG-positive foods because of an emotional reaction to restriction, and thus subsequently increased desire for the food(s) that was not overcome by dietary counseling.

This study was limited by the small sample size, which may have contributed to the lack of significant findings. Dietary intake patterns were not controlled as a part of this study and were self-reported by participants. Additionally, investigators did not control for the effects of exercise, stress management interventions, and other factors related to systemic inflammation or weight loss on the results of this study. There was no limitation placed on the waitlist group preventing them from making their own dietary changes or seeking other interventions during their wait period. The waitlist group was not aware of their targeted foods, but by chance the FFQ indicated they had reduced consumption of some of these foods, which may have impacted the IgG results. It is possible that participants in both the intervention and waitlist groups introduced other behavioral changes into their daily lives during the course of the study, which could have affected their PROMIS-29 and MYMOP scores. Another consideration is that both the number and type of targeted foods varied among participants. For analysis purposes, sensitivities to different foods were averaged within the individual, effectively treating all food sensitivities as equal. It should also be taken into consideration that 3 months may not have been a sufficient enough time to demonstrate a reduction in IgG levels. IgG testing needs more experimental research applied to test its value, and to better understand if it is useful in clinical practice.

Despite these limitations and the lack of significant findings, the results of this study are relevant. On average, IgG antibody concentrations were reduced in 83% of the targeted foods in the intervention group compared with only 60% in the waitlist group. These results support the conceptual framework that reduction or elimination of a food can lead to a decreased IgG concentration over time. The present study shows preliminary evidence of interest for this kind of research and illuminates the kind of outcomes that can be expected with the clinical implementation of elimination diets.

Footnotes

Acknowledgments

The authors thank Leslie Fuller, ND, for her help with planning of the project. They also thank Constance Ohlinger and Mary Parker for their help with the data collection and phone calls. Alletess Medical Laboratory provided all the testing for free as part of this study. Alletess also provided a stipend for research time during the first year of work.

Author Disclosure Statement

There are no conflicts of interest with any of the authors.

References

Schäfer

, Böhler

, Ruhdorfer

, et al. Epidemiology of food allergy/food intolerance in adults: Associations with other manifestations of atopy. Allergy Eur J Allergy Clin Immunol, 2001; 56:1172–1179.

Teufel

, Biedermann

, Rapps

, et al. Psychological burden of food allergy. World J Gastroenterol, 2007; 13:3456–3465.

Verrill

, Bruns

, Luccioli

. Prevalence of self reported food allergy in US adults: 2001, 2006, and 2010. Allergy Asthma Proc, 2015; 36:458–467.

Bischoff

, Crowe

. Gastrointestinal food allergy: New insights into pathophysiology and clinical perspectives. Gastroenterology, 2005; 128:1089–1113.

Gaby

. The role of hidden food allergy/intolerance in chronic disease. Altern Med Rev, 1998; 3:90–100.

Gocki

, Bartuzi

. Role of immunoglobulin G antibodies in diagnosis of food allergy. Postepy Dermatol Alergol, 2016; 33:253–256.

Vojdani

. Molecular mimicry as a mechanism for food immune reactivities and autoimmunity. Altern Ther Health Med, 2015; 21:34–45.

Yum

, Pyun

. Oral food challenges in children. Korean J Pediatr, 2011; 54:6–10.

Overweight and Obesity Statistics, 2012. Online document at: www.niddk.nih.gov, accessed May 5, 2018 .

10.

Bhaskaran

, Douglas

, Forbes

, et al. Body-mass index and risk of 22 specific cancers: A population-based cohort study of 5 · 24 million UK adults. Lancet, 2014; 384:755–765.

11.

Martin-Rodriguez

, Guillen-Grima

, Martí

, Brugos-Larumbe

. Comorbidity associated with obesity in a large population: The APNA study. Obes Res Clin Pract, 2015; 9:435–447.

12.

Greenberg

, Obin

. Obesity and the role of adipose tissue in inflammation and metabolism. Am J Clin Nutr, 2006; 83:461–465.

13.

Berin

, Mayer

. Immunophysiology of experimental food allergy. Mucosal Immunol, 2009; 2:24–32.

14.

Fink

. Intestinal epithelial hyperpermeability: Update on the pathogenesis of gut mucosal barrier dysfunction in critical illness. Curr Opin Crit Care, 2003; 9:143–151.

15.

Wilders-Truschnig

, Mangge

, Lieners

, et al. IgG antibodies against food antigens are correlated with inflammation and intima media thickness in obese juveniles. Exp Clin Endocrinol Diabetes, 2008; 116:241–245.

16.

Lewis

, Woolger

, Melillo

, et al. Eliminating immunologically-reactive foods from the diet and its errect on body composition and quality of life in overweight persons. J Obes Weight Loss Ther, 2012; 2:2–7.

17.

Ryan

, Heaner

. Guidelines (2013) for managing overweight and obesity in adults. Preface to the full report. Obesity (Silver Spring), 2014; 22:S1–S3.

18.

Appel

, Clark

, Yeh

H-C

, et al. Comparative effectiveness of weight-loss interventions in clinical practice. N Engl J Med, 2011; 365:1959–1968.

19.

Burke

, Dunbar-Jacob

. Adherence to medication, diet, and activity recommendations: From assessment to maintenance. J Cardiovasc Nurs, 1995; 9:62–79.

20.

Jathar

. Assessment of food specific immunoglobulin load in pre-school children. World Allergy Organ, 2015; 8(Suppl 1):A54.

21.

National Health and Nutrition Examination Survey. Center for Disease Control and Prevention. Online document at: www.cdc.gov/nchs/nhanes/about_nhanes.htm, accessed March 17, 2018 .

22.

Paterson

. Measuring outcomes in primary care: A patient generated measure, MYMOP, compared with the SF-36 health survey. BMJ, 1996; 312:1016–1020.

23.

Paterson

, Britten

. In pursuit of patient-centred outcomes: A qualitative evaluation of the “Measure Yourself Medical Outcome Profile.”. J Heal Serv Res Policy, 2000; 5:27–36.

24.

Marshall

, Altman

, Holder

, Royston

. Combining estimates of interest in prognostic modelling studies after multiple imputation: Current practice and guidelines. BMC Med Res Methodol, 2009; 9:57.

25.

Greenway

. Physiological adaptations to weight loss and factors favouring weight regain. Int J Obes, 2015; 39:1188–1196.

26.

Ogden

, Karim

, Choudry

, Brown

. Understanding successful behaviour change: The role of intentions, attitudes to the target and motivations and the example of diet. Health Educ Res, 2007; 22:397–405.

27.

Fedoroff

, Polivy

, Herman

. The effect of pre-exposure to food cues on the eating behavior of restrained and unrestrained eaters. Appetite, 1997; 28:33–47.