History of Trying Exclusion Diets and Association with Avoidant/Restrictive Food Intake Disorder in Neurogastroenterology Patients: A Retrospective Chart Review

Abstract

Background:

Exclusion diets for gastrointestinal symptom management have been hypothesized to be a risk factor for avoidant/restrictive food intake disorder (ARFID; a non-body image-based eating disorder). In a retrospective study of pediatric and adult neurogastroenterology patients, we aimed to (1) identify the prevalence and characteristics of an exclusion diet history and (2) evaluate if an exclusion diet history was concurrently associated with presence of ARFID symptoms.

Methods:

We conducted a chart review of 539 consecutive referrals (ages 6-90, 69% female) to adult (n=410; January-December 2016) and pediatric (n=129; January 2016-December 2018) neurogastroenterology clinics. Blinded coders (n=4) retrospectively applied DSM-5 criteria for ARFID and a separate coder assessed documentation of exclusion diet history. We excluded patients with no documentation of diet in the chart (n=35) or who were not orally fed (n=9).

Results:

Of 495 patients included, 194 (39%) had an exclusion diet history, and 118 (24%) had symptoms of ARFID. Of reported diets, dairy-free was the most frequent (45%), followed by gluten-free (36%). Where documented, exclusion diets were self-initiated by patients/parents in 66% of cases, and recommended by gastroenterology providers in 30%. Exclusion diet history was significantly associated with the presence of ARFID symptoms (OR=3.12[95% CI 1.92–5.14], p<.001).

Conclusions:

History of following an exclusion diet was common and was most often patient-initiated among pediatric and adult neurogastroenterology patients. As patients with self-reported exclusion diet history were over three times as likely to have ARFID symptoms, providers should be cognizant of this potential association when considering dietary interventions.

Graphical Abstract

INTRODUCTION

Disorders of gut-brain interaction (DGBI; also known as functional gastrointestinal (GI) disorders) encompass an array of sensory and motility disturbances affecting both pediatric and adult patients.^{(1, 2)} Exclusion diets (e.g., dairy-free diet, gluten-free diet) are often used for symptom management in DGBI, but may put some patients at risk for impaired nutrition and/or quality of life.^{(3, 4)}

Dietary management for DGBI is increasingly employed as a first-line intervention or as part of integrated care.^(5-7) Exclusion diets may be self-initiated by patients or selected by patients and providers due to their high face validity as a “natural,” non-pharmacologic treatment strategy.^{(3, 8)} In a survey of >1000 gastroenterologists in the United States, over half of providers recommended dietary therapy to adult patients with irritable bowel syndrome (IBS).⁽⁹⁾ Avoidance of specific foods is also common in patients with other DBGIs,⁽¹⁰⁾ such as functional dyspepsia,^{(11, 12)} and among pediatric patients with DGBI.⁽¹³⁾ However, with the increased use and popularity of exclusion diets, there is growing concern for their potential psychological impact and relationship to disordered eating.^{(3, 14, 15)}

High rates of avoidant/restrictive food intake disorder (ARFID) symptoms in DGBI^(16-20) have raised concern that exclusion diets may put some patients at risk for impairments in nutritional status or eating-related quality of life.^{(3, 8, 21)} ARFID is a non-body image-based eating disorder characterized by a reduction in food intake (volume and/or variety) associated with medical (e.g., inability to gain weight, nutrient deficiency) and/or psychosocial (e.g., social eating difficulty) impairments. ARFID symptoms have been reported in 13-40% of patients with DGBI.^{(8, 17-20)} Exclusion diets have been proposed as one potential risk factor for some patients to develop ARFID,^{(3, 8, 15, 22)} but whether this relationship exists remains to be elucidated.

In a retrospective study of pediatric and adult patients presenting for neurogastroenterology evaluation, we aimed to: (1) identify the prevalence and characteristics of current or past exclusion diet use and (2) evaluate if a history of exclusion diet was associated with presence of ARFID symptoms. We hypothesized that patients with a history of exclusion diet would have a significantly increased likelihood of having ARFID symptoms. We also explored the relationships between presence of ARFID symptoms and exclusion diet type (e.g., diary-free diet, gluten-free diet), and recommender (e.g., GI provider, self).

METHODS

Procedures

We conducted a retrospective chart review of patients (N=539, ages 6 – 90 years, 69% female) presenting to our tertiary care center’s pediatric and adult neurogastroenterology clinics. Pediatric patients (n=129, ages 6 – 18, 56% female) comprised consecutive referrals to the pediatric neurogastroenterology clinic from January 2016 to December 2018, and adult patients (n=410, ages 18 – 90, 73% female) comprised consecutive referrals to the adult neurogastroenterology clinic from January through December 2016. Pediatric referrals were collected over a longer timeframe (3 years versus 1 year for adult referrals) to maximize the sample size because of the lower patient volume in the pediatric neurogastroenterology clinic compared to the adult neurogastroenterology clinic.

We reviewed electronic medical records beginning at the initial consultation note and obtained demographic and clinical characteristics, including age and BMI at visit, sex, gender identity, race, ethnicity, presenting complaints, and gastrointestinal diagnoses conferred by providers. Masked coders (HBM, CJS, APB, FUR, CB) completed a diagnostic checklist, based on the Diagnostic and Statistical Manual for Mental Disorders, 5^th edition (DSM-5)⁽²³⁾ ARFID criteria, to indicate presence or absence of ARFID with moderate to substantial agreement (κ=0.60 for pediatric patients, κ=0.66 for adult patients). Coders conferred “definite ARFID” when cases met all DSM-5 ARFID criteria and “potential ARFID” when criteria were present but not enough information was available to confer or rule out a full diagnosis.^{(18, 19)} These groups were combined and categorized as “ARFID symptoms.”

A separate coder (MA) reviewed each electronic medical record from the initial consultation as well as all preceding hospital and clinic visits for documentation of diet history. Presence of exclusion diet was identified based on patient self-report of historical or ongoing exclusion diet in the medical records. Patients were not considered to have presence of exclusion diet if they had no documentation of history of trying an exclusion diet (i.e., trying a diet prior to initial gastroenterology consultation); this included patients who were advised to initiate an exclusion diet at the initial gastroenterology consultation or subsequent visits. We categorized exclusion diets based on five primary eliminated food groups or ingredients– diary-free, gluten-free, the low fermentable oligosaccharides, disaccharides, monosaccharides and polyols (FODMAP) diet, low acid diet, and low fat diet. We selected these diet categories as they had been previously described and studied in DGBI^{(6, 12)} and are frequently employed in clinical practice. Exclusion diets which did not fit in the major categories were categorized as “other exclusion diets” (e.g., low histamine, “Candida diet”). When noted by in the medical record, we categorized whether exclusion diet was self-/parent-initiated or provider recommended by patient report. Provider categories were comprised of GI provider, dietitian and “other provider” (e.g., primary care physician, naturopathic provider). We excluded patients from analysis if there was no documentation of diet in the medical record (n=35) or if they received parenteral nutrition or tube feeds (n=9). We did not classify patients as having an exclusion diet if they had: documented Celiac disease and were restricting only gluten, documented food allergies who were restricting only allergic triggers, or documented lactose intolerance and were only restricting dairy.

Institutional Review Board approval (Protocol 2018P002539 and 2018P000504) was obtained prior to study procedure commencement.

Statistical Analysis

We calculated frequency of exclusion diets, types of exclusion diets, and who recommended exclusion diets. We reported descriptive statistics to compare those with and without exclusion diets. For exploratory purposes, we compared frequency of exclusion diet type and who recommended exclusion diet among those with versus without ARFID symptoms. For all descriptive comparisons, we used Kruskal Wallis H-Tests for continuous data (as age and BMI were non-normally distributed) and chi-square tests for categorical data.

To test the hypothesis that history of exclusion diet would be associated with presence of ARFID symptoms, we conducted a multivariate logistic regression with presenting complaints as a priori covariates based on clinical relevance and demographic covariates selected based on univariate screen. We selected presenting complaints as covariates to control for the significant relations previously found between GI presentation and ARFID status.^{(18, 19)}Analyses were performed using R programming (R Core Team 2021). For sensitivity analysis, we also ran the regression models with the pediatric clinic sample and the adult clinic samples separated.

RESULTS

Table 1 includes demographic characteristics of the sample, presenting complaints, and gastrointestinal diagnosis, separated by presence of an exclusion diet history.

Table 1.

Characteristics of neurogastroenterology patients by exclusion diet history (N= 495).

	Exclusion Diet History (n=194)	No Exclusion Diet History (n=301)	p-valuea
Age, mean (SD)	38.1 (20.5)	39.1 (22.7)	0.733
Age group, n (%)
6 – 17 years	37 (19%)	75 (25%)	--
18 – 24 years	25 (13%)	19 (8%)	--
25 – 39 years	44 (23%)	57 (19%)	--
40 – 64 years	63 (32%)	98 (33%)	--
65 – 84 years	24 (12%)	43 (14%)	--
85+ years	1 (1%)	4 (1%)	--
Pediatric Clinic Patient, n (%)	40 (21%)	79 (26%)	--
Sex—female, n (%)	158 (81%)	184 (61%)	<0.001
Gender Identity—n (%)b
Female	156 (80%)	179 (59%)	<0.001
Male	36 (19%)	119 (40%)	--
Non-binary	2 (1%)	1 (0.3%)
Race, n (%)c
American Indian or Alaskan Native	1 (0.5%)	1 (0.3%)	--
Asian	5 (3%)	16 (5%)	--
Black or African American	5 (3%)	14 (5%)	--
Native Hawaiian or Pacific Islander	0 (0%)	0 (0%)	--
White	176 (91%)	238 (80%)	--
Other/Unknown	7 (4%)	27 (9%)	--
Ethnicity—Hispanic/Latinx, n (%)d	10 (5%)	26 (9%)	--
BMI, mean (SD)e	24.2 (6.2)	24.9 (6.6)	0.138
BMI, mean (SD) – patients > 18	25.4 (6.1)	26.6 (5.9)	0.018
BMI percentile, mean (SD) – patients < 18	54.5 (31.5)	57.8 (34.3)	0.432
ARFID symptoms, n (%)	73 (38%)	45 (15%)	<0.001
Presenting Complaints, n (%)f
Eating-related complaint	37 (19%)	41 (14%)	0.134
Esophageal-related complaint	75 (39%)	131 (44%)	0.328
Stomach-related complaint	80 (41%)	124 (41%)	1
Lower GI-related complaint	144 (74%)	182 (60%)	0.002
Gastrointestinal Diagnosis, n (%)f
Esophageal Diagnosis	59 (30%)	126 (42%)	0.013
Dyspepsia/Nausea/Vomiting Diagnosis	67 (35%)	69 (23%)	0.006
Lower GI Diagnosis	131 (68%)	162 (54%)	0.003
Chronic Abdominal Pain Diagnosis	19 (10%)	29 (10%)	1

Note. ARFID=avoidant/restrictive food intake disorder; BMI=body mass index; GI=gastrointestinal. Eating-related complaints included low weight, weight loss, poor appetite, and food aversion. Esophageal complaints included reflux, regurgitation, belching, globus, choking, cough, dysphagia, heartburn, chest pain/pressure, and other throat-related symptoms. Stomach complaints included nausea, vomiting, post-prandial fullness, early satiety, epigastric pain, and upper abdominal pain. Lower GI complaints included lower abdominal pain, diarrhea, constipation, fecal incontinence, rectal urgency, bloating/distension, rectal pain, anal itching, and flatulence. Esophageal diagnoses include globus, dysphagia, non-cardiac chest pain, gastroesophageal reflux disease, rumination syndrome, and chronic belching. Dyspepsia/nausea/vomiting diagnoses include gastroparesis, functional dyspepsia, cyclic vomiting syndrome, and chronic unspecified nausea and vomiting. Lower GI diagnoses include irritable bowel syndrome, chronic constipation, functional diarrhea, and abdominal bloating/distension.

^aComparisons presented for descriptive purposes. Continuous variables analyzed with Kruskal Wallis H-Tests for non-normally distributed data. Categorical variables analyzed with Chi-square Tests.

^bGender identity data missing for n=2. Gender identity categories included female, male and non-binary.

^cRace data missing for n=3.

^dEthnicity data missing for n =12.

^eBMI data missing for n=21.

^fn=214 had >1 symptom complaint. n=191 had >1 gastrointestinal diagnosis.

Prevalence and Characteristics of Exclusion Diets

Exclusion diet history was present in 194 patients (39%, 194/495), including 40 patients (34% 40/119) in the pediatric neurogastroenterology clinic sample, and 154 cases (41% 154/376) in the adult neurogastroenterology clinic sample. Patients with a history of exclusion diet were more likely to be female (81% vs. 61%, p<0.001), and, in the adult clinic patients only, to have lower mean BMI (p=0.018; but still in “normal range,” on average) (Table 1). Patients with a history of exclusion diet were also more likely to present with a lower gastrointestinal-related complaint (74% vs. 60%, p=0.002) and have a lower gastrointestinal or dyspepsia/nausea/vomiting diagnosis (68% vs. 54%, p=0.003 and 35% vs. 23%, p=0.006, respectively). Esophageal diagnoses were less frequent among patients with history of exclusion diets (30% vs. 42%, p=0.013) (Table 1).

Table 2 displays the frequencies of exclusion diet types. Among patients with exclusion diets in both the pediatric (n=40) and adult clinic (n=154) groups, dairy-free diets were the most frequent exclusion diet (45%, 87/194), although more frequent in the pediatric group (78%, 31/40) than in the adult group (36%, 56/154). In both patient groups, gluten-free diets were the second most frequent exclusion diet (40%, 16/40 in the pediatric group; 35%, 54/154 in the adult group).

Table 2.

Type of Historical Exclusion Diets Among Patients Presenting for Neurogastroenterology Consultation by ARFID Status (n = 194).

Type of Exclusion Diet, n (%)a	Total		ARFID symptoms (n = 73)	No ARFID symptoms (n = 121)	X2	p-valueg
Dairy-freea	87 (45%)		33 (45%)	54 (45%)	<0.001h	1
Gluten-freeb	70 (36%)		25 (34%)	45 (37%)	0.067	0.795
Low FODMAP dietc	53 (27%)		23 (32%)	30 (25%)	0.732	0.395
Low acid dietd	20 (10%)		2 (3%)	18 (15%)	6.000	0.014
Low fat diete	15 (8%)		9 (12%)	6 (5%)	2.511	0.113
Other exclusion dietf	55 (28%)		29 (41%)	26 (21%)	6.585	0.010
Multiple Exclusion Diets	80 (41%)		33 (45%)	47 (39%)	0.521	0.470

Note. ARFID=avoidant/restrictive food intake disorder; FODMAP= fermentable oligosaccharides, disaccharides, monosaccharides and polyols. The total sample (n=194) included adult (n=154) and pediatric (n=40) patients who had a history of an exclusion diet.

^an=31 pediatric patients, n=56 adult patients.

^bn=16 pediatric patients, n=54 adult patients.

^cn=6 pediatric patients, n=47 adult patients.

^dn=1 pediatric patients, n=19 adult patients.

^en=1 pediatric patients, n=14 adult patients.

^fn=14 pediatric patients, n=41 adult patients.

^gAnalyzed with Chi-square Tests for descriptive purposes.

^hExact value = 7.5651x10⁻³¹

Of 194 patients with exclusion diet history, 169 (n=38 pediatric clinic patients, n=131 adult clinic patients) had documented information about who recommended the exclusion diet (n=25 had no documentation of recommender). Most exclusion diets were self- or parent-initiated (66%, 112/169), but more frequently so in adult clinic patients (70%, 92/131) than in the pediatric clinic patients (53%, 20/38) (Table 3). Among the pediatric clinic group in particular, exclusion diets were also frequently recommended by providers, with 45% of diets (17/38) recommended by a GI provider and 13% of diets (5/38) recommended by a dietitian. Of note, there was overlap within exclusion diet history, with some patients having more than one type of exclusion diet (n=81), and/or diets recommended by multiple providers (e.g., recommended by GI provider and dietitian). Some diets were also self/parent-initiated and provider recommended (n=23).

Table 3.

Who Recommended Exclusion Diet Among Patients Presenting for Neurogastroenterology Consultation by ARFID Status (n = 169).^a

Who Recommended Exclusion Diet, n (%)b	Total		ARFID symptoms (n = 70)	No ARFID symptoms (n = 99)	X2	p-valuec
Self-Initiatedd	112 (66%)		54 (77%)	58 (59%)	5.514	0.019
GI Provider Recommendede	51 (30%)		19 (27%)	32 (32%)	0.305	0.581
Other Recommendedf	20 (12%)		5 (7%)	15 (15%)	1.812	0.178
Dietitian Recommendedg	11 (7%)		6 (9%)	5 (5%)	0.357	0.550

Note. ARFID=avoidant/restrictive food intake disorder; GI=gastrointestinal.

^an=25 did not have documentation of who recommended dietary restriction.

^bn=23 had exclusion diet recommended by > 1 provider.

^cAnalyzed with Chi-square Tests for descriptive purposes.

^dn=20 pediatric patients, n=92 adult patients

^en=17 pediatric patients, n=34 adult patients

^fn=3 pediatric patients, n=17 adult patients

^gn=5 pediatric patents, n=6 adult patients

Association Between Exclusion Diets and ARFID Symptoms

Univariate comparisons showed ARFID symptoms were significantly more frequent in patients with exclusion diet history compared to patients without exclusion diet history (38% vs 15%; Table 2). In the pediatric clinic group, 30 (25%, 30/119) had ARFID symptoms, with the following frequencies for motivations behind avoidant/restrictive eating: 17 (47%) fear of aversive consequences, 10 (33%) lack of interest/low appetite, and 3 (10%) sensory sensitivity [n=7 had multiple presentations]. In the adult clinic group, 88 (23%, 88/376) had ARFID symptoms, with the following frequencies for motivations behind avoidant/restrictive eating: 81 (92%) fear of aversive consequences, 20 (23%) lack of interest/low appetite, and 0 (0%) sensory sensitivity [n=13 had multiple presentations].

Logistic regression (Table 4) confirmed that exclusion diet history was significantly associated with the presence of ARFID symptoms (OR=3.12, 95% CI 1.92–5.14, p<0.001). This association remained similar when the model was run with the pediatric clinic group alone (OR=5.74, 95% CI 1.88—19.33, p=0.003) and the adult clinic group alone (OR=2.86, 95% CI 1.59—5.26, p<0.001). Of the covariates included, female sex, BMI, presenting with eating-related complaints, or presenting with stomach-related complaints were also significant predictors of ARFID symptom presence.

Table 4.

Multivariate Analysis of the Association of Exclusion Diet History with ARFID symptom presence.^a

	OR	95% CI	p-value
Exclusion diet history, yes/no	3.12	1.92 – 5.14	<0.001
Age	1	0.99 – 1.01	0.732
Female sex, yes/no	1.90	1.06 – 3.51	0.035
BMI	0.94	0.90 – 0.98	0.007
Eating-related complaints, yes/no	4.39	2.47 – 7.89	<0.001
Esophageal-related complaints, yes/no	0.70	0.41 – 1.18	0.182
Stomach-related complaints, yes/no	1.78	1.08 – 2.92	0.023
Lower GI-related complaints, yes/no	0.99	0.58 – 1.70	0.975

Note. Presenting complaints were selected as a priori covariates. Other variables (age, sex, BMI) were selected based on univariate screen (i.e., they were significantly different between those with and without ARFID symptoms—see Table 1).

^aModel provided a good fit [X²(8)=91.98, p<0.0001].

In exploratory comparisons, patients with ARFID symptoms were more likely to report history of self-initiated diet (77% vs. 59%, p=0.023), while rates of other provider-recommended diets did not significantly differ between those with and without ARFID symptoms (Table 3). Having a history of an “other exclusion diet” was significantly associated with having ARFID symptoms, while history of low acid diet was significantly associated with not having ARFID symptoms.

DISCUSSION

While food avoidance and exclusion diet use can be adaptive and helpful for symptom management for some patients, there is growing concern that exclusion diets may put a subset of patients at risk for ARFID.^{(3, 4, 8, 22)} In this retrospective study of pediatric and adult patients with DGBI, we found that more than one third (39%) of patients had a history of following an exclusion diet, most of which were self-initiated (66%), with dairy-free diets being most common (45%). Importantly, we found that self-reported exclusion diet history was significantly associated with ARFID symptom presence, which has implications for screening and management in DGBI.

Our findings on the frequency and characteristics of exclusion diet use add to the literature in DGBI, and to our knowledge is the first to report use of specific exclusion diets in a pediatric DGBI population. Patients with DGBI commonly try to identify dietary causes for symptoms. In fact, up to 90% of children/adolescents^{(24, 25)} and adults^{(26, 27)} with IBS attribute their symptoms to food. In patients of all ages, foods rich in carbohydrates and fat are most frequently indicated as symptom-producing.^{(24, 25, 28)} Thus, dairy-free, gluten-free and the low FODMAP diets have been embraced as treatment strategies in DGBI,^{(29, 30)} and were the most commonly reported exclusion diets in our study.

The relatively high rate of history of exclusion diet use is notable, especially given the limited evidence for many of the diets reported in our sample. The most rigorously studied exclusion diet in DGBI is the low FODMAP diet,^(31-34) but its efficacy is limited to adult patients with IBS.^{(31, 35)} There is minimal evidence for gluten-free^{(31, 36, 37)} or dairy-free diets in IBS,⁽³⁸⁾ and evidence supporting exclusion diets in other DGBIs, such as a low fat diet for functional dyspepsia, is scarce.^{(39, 40)} In pediatrics, very few studies have addressed exclusion diets in DGBI,^(41-44) and the rare, small randomized controlled trials show mixed results.^{(45, 46)} Given the minimal support for exclusion diets beyond the low FODMAP diet, the high frequency of exclusion diet history use is surprising, but may be explained by the fact that most diets were patient-, rather than provider-initiated.

Most exclusion diets were self-initiated by patients and/or their parents (66%). This is consistent with gastroenterologists reporting that over half of their patients with irritable bowel syndrome attempt to self-manage their symptoms with dietary interventions, most often using a “trial and error” approach.⁽⁹⁾ The frequency of self-initiated diets may also be due growing availability of online resources about dietary interventions and the increased production and marketing of specialty food products (e.g., gluten-free, vegan), but concerns remain about both the quality of this content⁽⁴⁷⁾ and lack of dietetic supervision.^{(48, 49)} We also found that almost half of exclusion diets had been documented as recommended by a provider, with 30% of all reported diets recommended by a GI provider. While gastroenterologist-prescribed exclusion diets may be a preferred, safer alternative to patient-initiated diets, patients may have difficultly applying these recommendations without closer dietitian guidance and supervision.^{(50, 51)} Gastroenterologists may also not be able to fully evaluate exclusion diet appropriateness—for example, a retrospective study in adults showed that over 70% of patients who screened positive for ARFID were recommended a low FODMAP diet by their GI provider.⁽⁵²⁾ In fact, exclusion diets may put a subset of patients at risk for developing de novo problems with eating, such as ARFID.

While prior studies have shown that disordered eating can be found in diet-controlled chronic disease, including DGBIs,^{(20, 53-55)} our study is the first to our knowledge to examine exclusion diets as a retrospective correlate for ARFID. Our finding that patients with exclusion diet history were over three times as likely to have ARFID symptoms raises concerns about additional potential adverse outcomes of dietary management strategies. Exclusion diets, including a gluten-free diet and the low FODMAP diet, have been linked to poor diet quality and inadequate nutrient intake.^{(4, 49, 56, 57)} Although studies have been conflicting,⁽⁵⁸⁾ in some, low FODMAP diet resulted in decreased caloric intake^{(32, 59)} and weight loss.⁽⁶⁰⁾ Exclusion diets can also negatively affect quality of life (e.g., difficulty eating outside the home).^{(58, 61-63)} It is possible that the association we found between ARFID and exclusion diets exists because both are independent markers of DGBI disease severity. However, when exclusion diets are associated with medical/nutritional and/or quality of life impairments, we argue this indicates that the patient’s current dietary restrictions are a problem, and likely meet criteria for ARFID.

The reason for the strong association between exclusion diet history and ARFID symptoms is not known. Some patients with DGBI may perceive some symptom improvement with an exclusion diet, but anxiety around their symptoms returning can lead to extreme or prolonged dietary restrictions and dysfunctional illness beliefs, such as overestimation of negative consequences of expanding their diet.^{(15, 64)} This maladaptive response to exclusion diets can spiral into a negative feedback loop, as the nutritional and psychological consequences of ARFID may promote sensory and motility disturbances, thus worsening DGBI symptoms.^{(15, 65)}

It is also possible that exclusion diet history and ARIFD symptoms had a strong association because of inadequate supervision. Our exploratory findings showed that self-initiated, but not provider-recommended, exclusion diets were significantly more frequent in those with ARFID symptoms. Unsupervised patients may lack clear expectations for exclusion diets and timelines for re-introduction of food. For the low FODMAP diet in particular, dietitian guidance has been identified as an important factor to reduce risk of negative outcomes,⁽⁴⁹⁾ with one study showing that patients who were not seen by dietitian were less able to follow the three phases of the diet, and over one-fourth of patients followed the restrictive phase of the diet for longer than recommended.⁽⁵¹⁾

While further prospective research is needed to understand how some patients develop ARFID in the context of DGBI, our study provides preliminary evidence that use of exclusion diets may be related to ARFID, and that ARFID may develop in patients attempting to follow diets without provider guidance. However, as we cannot demonstrate causality, it is possible that we found a relationship between exclusion diet history and ARFID because some patients may have had ARFID prior to their DGBI. DGBIs are common among patients with eating disorders,^{(66, 67)} including those with ARFID.⁽⁶⁸⁾ Patients who have established disordered eating practices when they are diagnosed with DGBI may be more likely to try exclusion diets. However, it is highly plausible that ARFID developed in the context of DGBI in our sample. Other research has shown that patients with diet-controlled health conditions are more likely to develop disordered eating after the onset of dietary therapy, ^{(54, 64)} perhaps due to the inherent focus on food intake with dietary regimens. ^{(53, 55, 64)} In patients with ARFID and DGBI, most patients cite a fear around gastrointestinal symptom consequences as a driver for their disordered eating^(18-20) and this, rather than other ARFID presentations (i.e., sensory sensitivity to food characteristics, low appetite/lack of interest in eating), was most common among our patients.

This study had several limitations related to its retrospective chart design, and specifically reliance on free-text documentation of diet in notes. Patient self-report of diet can be inaccurate,⁽⁶⁹⁾ and we were not able to evaluate important factors around exclusion diet use (e.g., the duration of diet, adherence to diet, and motivation for trialing dietary therapy). Exclusion diets in pediatric and adult populations differ because of parental involvement and potential growth and development implications, but we were not able to explore this due to our smaller sample of pediatric patients. Our categorization of ARFID symptom presence was also limited by the retrospective design, and there was not information to definitively confer an ARFID diagnosis for 83 cases of the combined ARFID symptom group. Finally, while we controlled for age and gastrointestinal presenting complaints in our multivariate analysis, our inability to examine differential risk by gastrointestinal diagnosis is a limitation.

Our findings are novel and enhance our understanding of exclusion diets among patients with DGBI, and should provoke more critical investigation of their use and relationship with ARFID. Exclusion diets may be helpful and/or non-problematic for many patients, but their use may be overly restrictive when not indicated or backed by evidence, particularly in pediatrics where the data supporting exclusion diets for DGBI is especially limited. Providers should consider screening patients’ dietary histories closely, being mindful of self-initiated diets and preexisting fear or anxiety around food intake (see Chey et al.⁽²¹⁾ and Burton Murray et al. for guidance).⁽⁷⁰⁾ Before recommending exclusion diets, providers should consider the psychological and nutritional effects, and consider the risk level of certain diets.⁽⁷⁰⁾ Ideally, exclusion diets should be implemented with dietitian guidance, and patients should be followed regularly to assess their response, and to determine when foods can be re-introduced. Future longitudinal research is needed, especially to identify what factors put some DGBI patients, but not others, at risk for ARFID development.