Gastrointestinal Symptom‐Specific Anxiety Predicts Positive SCOFF Screens in Adults With Hypermobile Ehlers‐Danlos Syndrome and Hypermobility Spectrum Disorder and May Overestimate Eating Disorder Risk

ABSTRACT

Background/Aims

The Sick, Control, One, Fat, Food (SCOFF) is a questionnaire that screens for risk of eating disorders. Some symptoms and psychological factors associated with disorders of gut–brain interaction (DGBI) may resemble the features of eating disorders, potentially overestimating the risk of eating disorders in populations with high DGBI prevalence such as those with Hypermobile Ehlers–Danlos syndrome/hypermobility spectrum disorder (hEDS/HSD). This study aimed to identify factors associated with screening positive on SCOFF among adult with hEDS/HSD to explore the questionnaire's suitability for this population.

Methods

A secondary analysis of data from 670 adults with hEDS/HSD recruited from Ehlers–Danlos Support UK and tertiary Neurogastroenterology. χ ² tests and logistic regression were used to examine factors associated with positive SCOFF screening result and positive responses to individual SCOFF questions.

Results

DGBI presence did not predict a positive SCOFF screen. However, high gastrointestinal symptom‐specific anxiety (GSA) (AOR: 3.19, p < 0.001), fear of aversive consequences from eating (AOR: 1.97, p = 0.013), and anxiety (AOR = 2.53, p < 0.001) were significant predictors. Notably, high GSA strongly predicted a positive response to questions two (Do you worry you have lost Control over how much you eat?) and five (Would you say that Food dominates your life?) (Q2 AOR = 2.58, p = 0.006, Q5 AOR = 5.98, p < 0.001).

Conclusions

In hEDS/HSD, high GSA strongly predicted a positive SCOFF screening outcome, suggesting that SCOFF's questions might detect behaviors associated with worries around GI symptoms rather than an eating disorder. Positive screens in this population should be interpreted with caution and accompanied by GSA‐focused assessments to enhance biopsychosocial management.

Key Points

• The SCOFF is a brief questionnaire used to detect eating disorders. The SCOFF might not perform accurately in patients with Hypermobile Ehlers–Danlos syndrome/hypermobility spectrum disorder where gastrointestinal symptoms are common as they can influence SCOFF screening results.

• In this study high gastrointestinal symptom‐specific anxiety and fear of negative consequences from eating predicted a high eating disorder risk.

• Clinicians should interpret SCOFF results carefully in this population as a positive eating disorder screen could reflect behaviors, concerns and consequences from gastrointestinal symptoms rather than an eating disorder.

Plain Language Summary

People with hypermobile Ehlers–Danlos syndrome/hypermobility spectrum disorder (hEDS/HSD) often experience gastrointestinal symptoms. The SCOFF is a short questionnaire commonly used to screen for eating disorders, however, it may not work perfectly in this group because gastrointestinal symptoms can affect how people answer its questions. In this study, we looked at factors that predicts SCOFF screening results and found that people with high anxiety about gastrointestinal symptoms and fears about the negative consequences of eating were more likely to have a positive SCOFF screen. This suggests that a positive SCOFF screen may reflect concerns about gastrointestinal symptoms rather than an eating disorder. Therefore, clinicians should interpret positive SCOFF results carefully in individuals with hEDS/HSD and gastrointestinal disorders.

Among adults with hypermobile Ehlers–Danlos syndrome/hypermobility spectrum disorder, gastrointestinal symptom‐specific anxiety predicted positive SCOFF screening outcomes. These findings raise the possibility that factors related to GI symptom anxiety may contribute to positive eating disorder screening results independent of eating disorder pathology.

1. Introduction

Disorders of gut–brain interaction (DGBI) are a group of conditions identified by a range of gastrointestinal symptoms that meet Rome IV diagnostic criteria [1], and are estimated to affect approximately 40% of the global population [2]. DGBI and eating disorders frequently co‐occur, with a reported DGBI prevalence ranging from 39% to 98% among individuals with eating disorders [3, 4, 5, 6]. Conversely, eating disorder risk in DGBI populations is estimated at 30% [7], with ranges between 16% and 55% depending on the screening method and the type of eating disorder [8]. Eating disorder screening in individuals with DGBI is essential for comprehensive patient care, given their aforementioned comorbidity. This becomes clinically significant as certain therapeutic strategies for DGBI involve the implementation of elimination diets, which have been suggested to increase the risk of developing or aggravating an eating disorder [6, 9]. Accordingly, using the sick, control, one, fat, and food (SCOFF) questionnaire has been recommended before commencing DGBI treatment [10, 11, 12, 13]. The SCOFF is a brief questionnaire (Table 1), developed to identify probable cases of eating disorders [14]. Since its development, SCOFF has been validated on mixed‐gendered populations [15, 16] and different nationalities [17, 18]. However, as recently highlighted by Soliman and colleagues [7], SCOFF has not been psychometrically validated in patients with DGBI.

TABLE 1.

The five questions of the Sick, Control, One, Fat, Food, questionnaire.

SCOFF questions a , b : Do you make yourself Sick because you feel uncomfortably full? Do you worry you have lost Control over how much you eat? Have you recently lost more than One stone in a 3‐month period? Do you believe yourself to be Fat when others say you are too thin? Would you say that Food dominates your life?

^a One point for every “yes”; a score of ≥ 2 indicates a likely case of anorexia nervosa or bulimia.

^b As presented in Morgan et al. [14].

We speculate that certain consequences of DGBI can resemble the features of eating disorders that SCOFF measures, such as weight loss associated with functional dyspepsia [19] or excessive thoughts about food resulting from dietary management of IBS symptoms [20]. Another psychological factor commonly observed in individuals with DGBI that may similarly influence SCOFF screening outcomes is gastrointestinal symptom‐specific anxiety (GSA) [21, 22], which has previously been associated with a positive eating disorder screening results [23]. Table 2 outlines the potential mechanisms by which DGBI and GSA may inflate SCOFF positive screening results by mimicking behaviors typically associated with eating disorder.

TABLE 2.

Rationale for how the (Sick, Control, One, Fat, Food) eating disorder questionnaire can misinterpret symptoms and behaviors of disorders of gut–brain interaction or gastrointestinal symptom‐specific anxiety as behaviors of eating disorders.

Eating disorders behaviors and cognitions measured by SCOFF a	Rationale	Supporting evidence
First question: self‐induced vomiting	Self‐induced vomiting could be a measure of reliving physical discomfort associated with DGBI symptoms, or a behavioral response to increased visceral anxiety, rather than a way of controlling body weight.	13% of adolescents with IBS report vomiting after eating to control symptoms [24].
		Patients with higher esophageal symptom‐specific anxiety are quicker at inducing vomiting to relieve dysphagia [25].
Second question: obsessive control of food intake	Careful food monitoring may reflect efforts to control the intake of symptom‐triggering foods, rather than total caloric intake.	84% of patients with IBS report at least one symptom‐triggering food item [26], and 97.9% of adolescents with IBS report sometimes avoiding a triggering food item to manage symptoms [24].
		Higher GI symptom specific anxiety is associated with self‐prescribed dietary restrictions, such as avoiding dairy products [27].
		71% of individuals with functional dyspepsia report restricting FODMAPs, wheat, gluten, and dairy [28].
Third question: weight loss (desire to be thin)	Weight loss in DGBI can result from the condition itself rather than intentional efforts driven by body image.	34% of patients with functional dyspepsia lost 5% of their body weight after the onset of dyspepsia symptoms [29], with early satiety, nausea, and vomiting, being associated with an average of 4 kg weight loss [19]
	Avoiding symptom triggering foods can lead to decreased energy intake resulting in weight loss that is unintentional rather than one driven by a desire to be thin.	High GI symptom specific anxiety in patients with IBS, is associated with fewer calorie and carbohydrates intake [30].
		Patients with IBS who engage in severe food avoidance consume significantly less energy than those who don't [31].
		Patients with functional dyspepsia have lower energy and fat intake in comparison to healthy controls [32].
Fourth question: weight/shape concerns	Bloating and visible distension in certain DGBI may cause bodily dissatisfaction that is unrelated to weight concerns.	Individuals with IBS have greater body dissatisfaction in comparison to healthy controls [33].
		High GI symptom specific anxiety is associated with a negative perception of one's abdomen, and a decreased self‐esteem [34].
Fifth question: food‐dominated thoughts	In certain DGBI constant thoughts about food is due to symptoms concerns rather than disordered eating behaviors.	Chumpitazi et al. [20] and Tuck et al. [13] have noted that this question could be an indicator of the effects of diet on DGBI symptoms rather than a disordered eating behavior.

Abbreviations: DGBI, disorders of gut‐brain interaction; GI, gastrointestinal; IBS, irritable bowel syndrome; SCOFF, (Sick, Control, One Stone, Fat, Food) eating disorder screening questionnaire.

^a Adopted from Morgan et al. [35].

Our study focuses on individuals with hypermobile Ehlers–Danlos syndrome/hypermobility spectrum disorder (hEDS/HSD), a connective tissue disorders with a broad range of health challenges [36]. DGBI are particularly prevalent in this population, with two large cohort studies reporting that approximately 98% of individuals with hEDS/HSD meet criteria for at least one DGBI [37, 38]. Moreover, theoretical speculations suggests that oral symptoms of hEDS/HSD, such as taste alterations, and chewing or swallowing difficulties may contribute to the development of an eating disorder [39, 40, 41, 42, 43, 44]. Baeza‐Velasco et al. [45] found that 30.3% of patients with hEDS were detected of having high risk of eating disorders using the SCOFF. However, the presence of DGBI was not assessed in that study. In DGBI samples, the rates of eating disorders risk is nearly identical (30.2%) [7], and only slightly lower (25.4%) in patients with both hEDS/HSD and DGBI [38].

Given the high prevalence of DGBI in hEDS/HSD [37, 38], and the potential of DGBI and its associated symptoms to influence SCOFF outcome (Table 2), the interpretation of SCOFF results in this population may be challenging. It remains unclear whether the SCOFF can effectively identify eating disorder risk in individuals with hEDS/HSD. This could overestimate the risk of eating disorders in individuals with hEDS/HSD, and lead to misallocation of clinical resources. It can also heighten patients' sense of being misunderstood when their symptoms are misattributed to an eating disorder; similar to the stigmatization experienced by patients with DGBI whose physical symptoms are dismissed as psychiatric in nature [46]. Therefore, the primary aim of this study is to identify factors that predict a positive SCOFF screening result in individuals with hEDS/HSD. A secondary aim was to examine factors that predict a “yes” answer for individual SCOFF questions to identify questions that may be inappropriate in this population. We hypothesized that DGBI, particularly functional dyspepsia, and GSA will predict a positive SCOFF screening result.

2. Methods

2.1. Study Design

A secondary analysis of cross‐sectional data collected as part of the CANDI‐hEDS study (Comprehensive Assessment of Nutrition and Dietary Intervention in patients with hEDS) [38]. The CANDI‐hEDS was approved by London—Westminster Research Ethics Committee (Rec ref.: 21/PR/0916). Recruitment spanned August 2021 and January 2022. Participants were recruited from tertiary neurogastroenterogy clinics at two London teaching hospitals and through online advertisements via Ehlers–Danlos Support UK. Participants completed a set of online questionnaires and were included in this analysis if they had completed the SCOFF questionnaire and had no missing data (see Figure S1). Detailed study design, recruitment and data collection methods are described elsewhere [38].

2.2. Questionnaires

2.2.1. SCOFF

SCOFF has five yes/no questions that screen for risk of eating disorders [14]. Each question answered with “yes” is given one score. A score of two or more was determined to be the cutoff value for a positive screen, as it has shown a pooled sensitivity of 86% in detecting eating disorders cases [47].

2.2.2. Demographics and Anthropometrics

Demographics and anthropometrics were collected using a structured questionnaire that asked about gender, age, weight, and height. Age categories were defined as: young adults (18–24 years), and adults (≥ 25 years). Gender was reported as “female,” “male,” or “other.” Body mass index (BMI) was categorized into: underweight (< 18.5 kg/m²), normal (18.5–24.9 kg/m²), and overweight/obese (≥ 25 kg/m²) [48].

2.2.3. Diet Behavior

Diet behavior was evaluated using a structured questionnaire.

• Meal and snack patterns: participants reported the number of meals and snacks they typically consumed per day. Response options for both items were 1, 2, 3, “varies” or “other”. For analyses, meal patterns were categorized into three groups: 3, 1–2 meals/day, and “varies” or “other.” Snack patterns were categorized into three groups: 1–2, 3 snacks/day, and “varies” or “other.”

• Number of elimination diets tried: participants indicated which elimination diets they had previously tried. Response options included low FODMAP, gluten‐free, dairy‐free, gluten‐ and dairy‐free, low histamine, Whole30, paleo, autoimmune paleo, and specific carbohydrate diet (SCD). For analyses: the total number of elimination diets chosen by each participant was calculated manually and categorized as ≤ 2 elimination diets or ≥ 3 elimination diets.

2.2.4. Rome IV Diagnostic Questionnaire for Adults

DGBI were identified using participants' responses to the Rome IV diagnostic criteria, which was administered directly within the study questionnaire using the validated Rome IV Diagnostic Questionnaire for Adults [49].

2.2.5. Gastrointestinal Symptoms Rating Scale‐Irritable Bowel Syndrome (GSRS‐IBS)

The GSRS‐IBS has five subscales to measures the intensity of five gastrointestinal symptoms (abdominal pain, reflux, indigestion, constipation, and diarrhea) and is scored using a seven‐point Likert scale [50]. For analyses, a score of 1–4 was determined as the cutoff for mild/moderate symptoms, while a score of 5–7 was used as the cutoff for severe/very severe symptoms.

2.2.6. Visceral Sensitivity Index (VSI)

GSA was assessed using the VSI. The VSI uses a six‐point Likert scale where 1 represents (strongly agree) and 6 represents (strongly disagree). Total VSI score range from 0 to 75 [51]. For the purposes of analyses, VSI scores were grouped into three categories; low GSA (scores of 0–24), moderate GSA (scores of 25–49), and high GSA (scores of 50–75).

2.2.7. Hospital Anxiety and Depression Scale (HADS)

Anxiety and depression were assessed using the HADS screen. For analyses, the recommended cutoff score of ≥ 11 were used to identify probable anxiety or depression [46, 52].

2.2.8. Patient Health Questionnaire (PHQ)

Somatic symptom sensitivity was assessed using the 12‐item PHQ [53]. For analyses, a score of 0–12 was used as the cutoff for low somatic symptoms severity, while a score of 13–24 was used as the cutoff for high somatic symptoms severity.

2.2.9. Nine Item Avoidant/Restrictive Food Intake Disorder Screen (NIAS)

The NIAS is a screening tool for avoidant/restrictive food intake disorder (ARFID) that is designed to measures the drivers of ARFID [54]. For analyses purposes, a positive ARFID screen was defined as a total NIAS score of ≥ 24. For the NIAS subscales, previously validated cutoff scores were used: ≥ 10 for “picky eating,” ≥ 10 for “fear of aversive consequences from eating,” and ≥ 9 for “low appetite (lack of interest in food)” [55, 56].

2.2.10. Comorbidities

Self‐reported comorbidities were collected via a structured questionnaire. Among the collected comorbidities, the study included two medical comorbidities: mast cell activation syndrome (MCAS) and postural tachycardia syndrome (PoTS), and two psychiatric comorbidities: anorexia nervosa (AN) and bulimia nervosa (BN).

2.3. Statistical Analysis

All statistical analyses were performed using IBM SPSS Statistics v.29.0.

2.3.1. Outcomes

Primary outcome: factors independently associated with a positive SCOFF's screen (SCOFF score ≥ 2).

Secondary outcome: factors independently associated with a positive response to each of SCOFF's five questions.

2.3.2. Analysis

Pearson's χ ² tests were performed to determine which of the following variables (age, gender, BMI, meal pattern, snack pattern, number of elimination diets tried, severity of GI symptoms, DGBI, GSA, anxiety, depression, somatic symptoms, a positive ARFID screen, PoTs, MCAS, AN, and BN) was associated with a positive SCOFF screen. All variables, regardless of statistical significance on the χ ² analysis, were then entered as covariates into a logistic regression model to identify independent predictors of overall SCOFF outcome and individual SCOFF items. A summary of how the data were prepared and handled during the analysis is detailed in Figure S1. A p value of < 0.05 was considered significant.

3. Results

3.1. Sample Characteristics

Six hundred and seventy participants were included (95% female), with a mean age of 40.5 (SD = 13.05; range 18–76 years). Almost half (47%) of participants had an overweight/obese BMI compared to 44% with a normal BMI, and only 8% with an underweight BMI (Table 3). Almost all participants (98%) fit Rome IV diagnostic criteria for one or more DGBI diagnoses, with the most prevalent disorders being functional dyspepsia (62%), IBS (45%), functional dysphagia (39%), and chronic nausea and vomiting syndrome (CNVS) (38%) (Table 3). Additionally, 25% (n = 173) of participants screened positive for high risk of eating disorders using the SCOFF questionnaire. Detailed sample characteristics are presented in Table 3.

TABLE 3.

Characteristics of study variables.

Variables	Categories	Total (n = 670)
		N	(%)
Age	Young adults (18–24 years old)	83	12.4
	Adult (≥ 25 years old)	587	87.6
Gender	Female	640	95.5
	Male	28	4.2
	Other	2	0.3
Body mass index	Underweight (≤ 18.4 kg/m2)	54	8.1
	Normal (18.5–24.9 kg/m2)	297	44.3
	Overweight/obese (≥ 25 kg/m2)	319	47.6
Screening for risk of eating disorders
Positive screen (SCOFF score ≥ 2)		173	25.8
Dietary behaviors
Meal pattern	3 meals/day	270	40.3
	1–2 meals/day	218	32.5
	Irregular	182	27.2
Snack pattern	1–2 snacks/day	315	47
	3 snacks/day	58	8.7
	Irregular	297	44.3
Elimination diets tried	≤ 2 elimination diets	397	59.3
	≥ 3 elimination diets	273	40.7
Disorders of gut–brain interaction a
At least one DGBI		661	98.6
Functional dyspepsia		417	62.2
Irritable bowel syndrome		305	45.5
Functional dysphagia		262	39.1
Chronic nausea and vomiting syndrome		260	38.8
Gastrointestinal symptom severity
Abdominal pain (GSRS‐IBS subscale, severe/very severe)		70	10.4
Reflux (GSRS‐IBS subscale, severe/very severe)		106	15.8
Indigestion (GSRS‐IBS subscale, severe/very severe)		104	15.5
Diarrhea (GSRS‐IBS subscale, severe/very severe)		118	17.6
Constipation (GSRS‐IBS subscale, severe/very severe)		151	22.5
Gastrointestinal symptom‐specific anxiety
Low GSA (VSI, 0–24)		187	27.9
Moderate GSA (VSI, 25–49)		322	48.1
High GSA (VSI, 50–75)		161	24
Anxiety and depression
Anxiety (HADS‐anxiety, ≥ 11)		287	42.8
Depression (HADS‐depression, ≥ 11)		129	19.3
Somatic symptom severity
High symptom severity (PHQ‐12, ≥ 13)		453	67.6
Avoidant/restrictive eating behaviors
Positive ARFID screen (NIAS, ≥ 24)		205	30.6
Picky eating (NIAS‐picky eating subscale, ≥ 12)		94	14
Fear of Aversive Consequences (NIAS‐fear subscale, ≥ 12)		213	31.8
Lack of interest in food (NIAS‐appetite subscale, ≥ 12)		249	37.2
Self‐reported comorbidities a
Postural orthostatic tachycardia syndrome		330	49.3
Mast cell activation syndrome		225	33.6
Anorexia nervosa		51	7.6
Bulimia nervosa		27	4

Abbreviations: ARFID, avoidant/restrictive food intake disorder; DGBI, disorders of gut‐brain interaction; GSA, gastrointestinal symptoms‐specific anxiety; GSRS‐IBS, gastrointestinal symptoms rating scale‐IBS; HADS, hospital anxiety and depression scale; NIAS, nine‐item avoidant/restrictive food intake disorder screen; PHQ, patient health questionnaire; SCOFF, Sick, Control, One Stone, Fat, Food eating disorder screening questionnaire; VSI, visceral sensitivity index.

^a The displayed DGBI reflect the four most prevalent in the study sample; comorbidities reflect the four conditions relevant to the study. Additional conditions are not shown.

3.2. Factors Associated With a Positive SCOFF Eating Disorders Screening Outcome

Pearson's χ ² analysis identified several factors that were significantly associated with a positive SCOFF eating disorder screening outcome (Table 4). Young adults (18–24 years) were more likely to screen positive than those aged ≥ 25 (42.1% vs. 23.5%, p < 0.001). BMI was also significantly associated with SCOFF screening outcomes (p = 0.004), with participants in the overweight/obese category having the highest prevalence of positive SCOFF screens (31.7%), compared with those who had an underweight (22.2%) or a normal BMI (20.2%). Among dietary behaviors, both meal pattern (p = 0.018), and snack pattern (p = 0.005), were significantly associated with SCOFF outcome. Participants consuming one or two meals per day had the highest prevalence of positive SCOFF screens (37.0%), compared with those reporting three meals per day (31.2%) or an irregular meal pattern (31.8%). Similarly, participants with an irregular snack pattern had the highest prevalence of positive SCOFF screens (53.2%), followed by those consuming one or two snacks per day (36.4%) and three snacks per day (10.4%).

TABLE 4.

Factors associated with a positive SCOFF screening outcomes.

Variable	Categories	Subjects with a positive SCOFF outcome		p
		N	(%)
Age	Young adults (18–24 years old)	35	42.1	< 0.001
	Adult (≥ 25 years old)	138	23.5
Gender	Female	167	26.1	0.161
	Male	4	14.3
Body mass index	Underweight	12	22.2	0.004
	Normal	60	20.2
	Overweight/obese	101	31.7
Dietary behavior
Meal pattern	3 meals/day	54	31.2	0.018
	1–2 meals/day	64	37
	Irregular	55	31.8
Snack pattern	1–2 snacks/day	63	36.4	0.005
	3 snacks/day	18	10.4
	Irregular	92	53.2
Elimination diets tried	≤ 2 elimination diet	113	28.5	0.059
	≥ 3 elimination diet	60	22
Gastrointestinal symptom severity
Abdominal pain	Mild/moderate	142	23.7
	Severe/very severe	31	44.3
Reflux	Mild/moderate	142	25.2	0.380
	Severe/very severe	31	29.2
Indigestion	Mild/moderate	134	23.7	0.003
	Severe/very severe	39	37.5
Diarrhea	Mild/moderate	131	23.7	0.008
	Severe/very severe	42	35.6
Constipation	Mild/moderate	130	25	0.397
	Severe/very severe	43	28.5
Disorders of gut–brain interaction a
Functional dyspepsia	Present	108	25.9	0.953
	Absent	65	25.7
Irritable bowel syndrome	Present	78	25.6	0.894
	Absent	95	26
Functional dysphagia	Present	72	27.5	0.431
	Absent	101	24.8
Chronic nausea and vomiting syndrome	Present	78	30	0.049
	Absent	95	23.2
Gastrointestinal symptom‐specific anxiety (GSA)
Low GSA		24	12.8	< 0.001
Moderate GSA		77	23.9
High GSA		72	44.7
Anxiety and depression
Anxiety	Present	114	39.7	< 0.001
	Absent	59	15.4
Depression	Present	60	46.5	< 0.001
	Absent	113	20.9
Somatic symptoms severity
High symptom severity	Present	131	28.9	0.008
	Absent	42	19.4
Avoidant/restrictive eating behaviors
Positive ARFID screen	Present	58	28.3	0.332
	Absent	115	24.7
Picky eating	Present	31	33	0.087
	Absent	142	24.7
Fear of aversive consequences	Present	71	33.3	0.002
	Absent	102	22.3
Lack of interest in food	Present	71	28.5	0.221
	Absent	102	24.2
Self‐reported comorbiditiesa
Postural orthostatic tachycardia syndrome	Present	84	25.5	0.831
	Absent	89	26.2
Mast cell activation syndrome	Present	54	24	0.444
	Absent	119	26.7
Anorexia nervosa	Present	20	39.2	0.023
	Absent	153	24.7%
Bulimia nervosa	Present	15	55.6	< 0.001
	Absent	158	24.6

Abbreviations: ARFID, avoidant/restrictive food intake disorder; GSRS, Gastrointestinal Symptoms Rating Scale; HADS, Hospital Anxiety and Depression Scale; NIAS, Nine‐Item Avoidant/Restrictive Food Intake Disorder Screen; PHQ, Patient Health Questionnaire; SCOFF, sick, control, one stone, fat, food eating disorder screening questionnaire; VSI, visceral sensitivity index.

^a The displayed DGBI reflect the four most prevalent in the study sample; comorbidities reflect the four conditions relevant to the study. Additional conditions are not shown.

p ≤ 0.05. Bold value indicate significant results.

Greater gastrointestinal symptom severity for abdominal pain (44.3% vs. 23.7%, p < 0.001), indigestion (37.5% vs. 23.7%, p = 0.003), and diarrhea (35.6% vs. 23.7%, p = 0.008), as well as high somatic symptoms severity measured by the PHQ‐12 (28.9% vs. 19.4%, p = 0.008), were significantly associated with a positive SCOFF screen. Participants with high GSA had a significantly higher prevalence of positive SCOFF screens (44.7%, p < 0.001) compared with those with moderate (23.9%) or low GSA (12.8%).

Meeting HADS criteria for anxiety (39.7% vs. 15.4%, p < 0.001), or depression (46.5% vs. 20.9%, p < 0.001), and meeting the NIAS's cutoff for “fear of aversive consequences” (33.3% vs. 22.3%, p = 0.002) were also significantly associated with a positive SCOFF outcome. Moreover, participants who self‐reported a diagnosis of AN (39.2% vs. 24.7%, p = 0.023) or BN (55.6% vs. 24.6, p < 0.001) had a significantly higher prevalence of positive SCOFF screens compared to their counterparts.

Among the analyzed DGBI, only CNVS was significantly associated with a positive SCOFF screen (30% vs. 23.2, p = 0.049), however, functional dyspepsia, functional dysphagia, IBS, self‐reported PoTS or MCAS were not significantly associated with a positive SCOFF outcome.

3.3. Predictors of a Positive SCOFF Eating Disorder Screen

A multivariable binary logistic regression was performed including all study variables regardless of their statistical significance in the univariate analyses (see Table S1 for the complete regression output). The analysis identified an overweight/obese BMI, an irregular snack pattern, high GSA, anxiety, depression, and fear of aversive consequences from eating as a statistically significant predictors of a positive SCOFF screening outcome. High GSA was the strongest independent predictor of a positive SCOFF screen, with participants classified as having high GSA demonstrating threefold higher odds of a positive SCOFF outcome than those with low GSA (AOR = 3.19, 95% CI: 1.59–6.45, p < 0.001). Anxiety was also a significant predictor, with participants meeting the HADS criteria for anxiety having more than twice the odds of a positive SCOFF outcome compared with those without anxiety (AOR = 2.53, 95% CI: 1.65–3.9, p < 0.001) (Table 5). Similarly, participants meeting the NIAS cutoff for “fear of aversive consequences” had nearly twice the odds of screening positive on the SCOFF (AOR = 1.97, 95% CI: 1.15–3.37, p = 0.013).

TABLE 5.

Binary logistic regression analysis for prediction of a positive SCOFF score at the cut‐off of ≥ 2.

Variables	AOR	95% CI		p
		Lower	Upper
Increased likelihood of positive response
Overweight/obese BMI	1.88	1.20	2.95	0.006
Irregular snack pattern	1.61	1.03	2.51	0.035
Anxiety	2.53	1.65	3.90	< 0.001
Depression	1.67	1.03	2.73	0.038
High GSA	3.19	1.58	6.45	< 0.001
Fear of aversive consequences	1.97	1.15	3.37	0.013
Decreased likelihood of positive response
Adult age group (≥ 25 years)	0.517	0.28	0.94	0.032
≥ 3 Elimination diets	0.601	0.38	0.94	0.027
Functional dyspepsia	0.599	0.36	0.99	0.047
Positive ARFID screen (NIAS, ≥ 24)	0.499	0.26	0.95	0.036

Abbreviations: AOR, adjusted odds ratio; ARFID, avoidant/restrictive food intake disorder; BMI, body mass index; CI, confidence interval; SCOFF, Sick, Control, One Stone, Fat, Food eating disorder screening questionnaire.

Additionally, the analysis revealed several factors associated with a decreased likelihood of a positive SCOFF screening outcome (Table 5). For instance, age was a significant predictor, with adults (≥ 25 years old) having nearly half the odds of a positive SCOFF result compared to young adults (18–24 years old) (AOR = 0.51, 95% CI: 0.28–0.95, p = 0.032). Furthermore, individuals who had tried three or more elimination diets were less likely to screen positive on the SCOFF compared to those who had tried fewer than three (AOR = 0.6, 95% CI: 0.38–0.94, p = 0.027). Similarly, participants with a positive ARFID screen had 50% lower odds of a positive SCOFF result compared to their counterparts (AOR = 0.49, 95% CI: 0.26–0.95, p = 0.036).

3.4. Predictors of Individual SCOFF Questions

To identify predictors of a positive response to each SCOFF item, binary multivariate logistic regression models were run separately for each question. Significant predictors are reported below with complete regression outputs provided in Tables S2–S6. Individuals who self‐reported BN had approximately six times higher odds of providing a positive response to SCOFF's first question “Do you make yourself sick because you feel uncomfortably full?” (AOR = 5.04, 95% CI: 1.44–17.59, p = 0.011). In contrast, respondents who had tried three or more elimination diets, or screened positive for ARFID using the NIAS questionnaire, exhibited approximately one‐third the odds of answering the question, compared to their counterparts (≥ 3 elimination diets tried; AOR = 0.33, 95% CI: 0.15–0.73, p = 0.006, positive ARFID AOR = 0.29, 95% CI: 0.10–0.82, p = 0.020) (Table 6).

TABLE 6.

Predictors of a positive response to individual SCOFF questions.

	AOR	95% CI		p
		Lower	Upper
Question 1: “Do you make yourself Sick because you feel uncomfortably full?”
Increased likelihood of positive response
Self‐reported bulimia nervosa	5.04	1.44	17.59	0.011
Decreased likelihood of positive response
≥ 3 Elimination diets	0.33	0.15	0.73	0.006
Positive ARFID screen	0.29	0.10	0.82	0.020
Question 2: “Do you worry you have lost Control over how much you eat?
Increased likelihood of positive response
Overweight/obese BMI	2.58	1.66	4.02	< 0.001
Irregular snack pattern	1.68	1.08	2.60	0.019
High GSA	2.58	1.30	5.12	0.006
Anxiety	1.89	1.24	2.88	0.003
Depression	2.40	1.48	3.88	< 0.001
Decreased likelihood of positive response
Functional dyspepsia	0.58	0.35	0.94	0.029
Question 3: “Have you recently lost more than One stone (6.4 kg) in a 3‐month period?”
Increased likelihood of positive response
Overweight/obese BMI	1.73	1.07	2.79	0.023
Having 1–2 meals/day	2.05	1.17	3.59	0.011
Irregular meal pattern	2.01	1.09	3.71	0.025
Anxiety	1.81	1.13	2.88	0.012
Fear of aversive consequences	2.23	1.24	4.01	0.007
Lack of interest in food	2.19	1.19	4.05	0.012
Question 4: “Do you believe yourself to be Fat when others say you are too thin?”
Increased likelihood of positive response
Having 1–2 meals/day	2.45	1.21	4.97	0.013
Anxiety	3.06	1.64	5.69	< 0.001
Self‐reported anorexia nervosa	4.12	1.75	9.69	< 0.001
Decreased likelihood of positive response
Adult age group (≥ 25 years)	0.43	0.20	0.89	0.023
Question 5: “Would you say that Food dominates your life?”
Increased likelihood of positive response
Female gender	4.91	1.04	23.10	0.044
Overweight/obese BMI	1.59	1.03	2.46	0.036
Irregular snack pattern	1.73	1.11	2.69	0.014
Moderate GSA	2.41	1.36	4.24	0.002
High GSA	5.98	2.94	12.15	< 0.001
Anxiety	1.73	1.14	2.64	0.010
Decreased likelihood of positive response
Functional dyspepsia	0.47	0.29	0.77	0.003
Positive ARFID screen	0.49	0.26	0.93	0.030

Note: Statistically significant variables only; full results that include statistically insignificant variables can be read in Table S2–S7. The displayed DGBI reflect the four most prevalent in the study sample; comorbidities reflect the four conditions relevant to the study. Additional conditions are not shown.

Abbreviations: AOR, adjusted odds ratio; ARFID, avoidant/restrictive food intake disorder; BMI, body mass index; CI, confidence interval; GSA, gastrointestinal symptom‐specific anxiety; SCOFF, Sick, Control, One Stone, Fat, Food eating disorder screening questionnaire.

The odds of answering “yes” to SCOFF's second question “Do you worry you have lost control over how much you eat?” were influenced by several factors. Participants with an overweight/obese BMI had higher odds of endorsing this item compared with those in other BMI categories (AOR = 2.58, 95% CI: 1.66–4.02, p < 0.001). Similarly, an irregular snacking pattern was associated with increased odds of a positive response (AOR = 1.68, 95% CI: 1.08–2.6, p = 0.019). Furthermore, high GSA (AOR = 2.58, 95% CI: 1.3–5.12, p = 0.006) anxiety (anxiety: AOR = 1.89, 95% CI: 1.24–2.88, p = 0.003), and depression (AOR = 2.4, 95% CI: 1.48–3.88, p < 0.001) were factors that predicted a positive response as well. In contrast, individuals with functional dyspepsia had a lower likelihood of giving a positive answer compared to their counterparts (AOR = 0.58, 95% CI: 0.35–0.94, p = 0.029).

Furthermore, the likelihood of providing a positive response to SCOFF's third question “Have you recently lost more than one stone in a 3‐month period?” was increased by factors such as having an overweight/obese BMI (AOR = 1.73, 95% CI: 1.07–2.79, p = 0.023), having 1–2 meals/day (AOR = 2.05, 95% CI: 1.17–3.59, p = 0.011), an irregular meal pattern (AOR = 2.01, 95% CI: 1.09–3.71, p = 0.025), anxiety (AOR = 1.81, 95% CI: 1.13–2.88, p = 0.012), meeting NIAS' cutoff score for the subscales “fear of aversive consequences” (AOR = 2.23, 95% CI: 1.24–4.01, p = 0.007) and “lack of interest in food” subscale (AOR = 2.19% CI: 1.19–4.05, p = 0.012).

A positive response to SCOFF's fourth question “Do you believe yourself to be fat when others say you are too thin?” was strongly predicted by self‐reported AN (AOR = 4.12, 95% CI: 1.75–9.69, p < 0.001), having 1–2 meals/day (AOR = 2.45, 95% CI: 1.21–4.97, p = 0.013), and anxiety (AOR = 3.06, 95% CI: 1.64–5.69, p < 0.001). Conversely, compared to younger adults (18–24 years), adults (≥ 25 years old) were less likely to answer affirmatively (AOR = 0.43, 95% CI: 0.2–0.89, p = 0.023).

Lastly, high GSA emerged as the strongest predictor for a positive response to SCOFF's fifth question “Would you say that food dominates your life?.” Individuals with high GSA had almost six times higher odds of answering “yes” in comparison to those with low GSA (AOR = 5.98, 95% CI: 2.94–12.15, p < 0.001). Other predictors included female gender (AOR = 4.91, 95% CI: 1.04–23.1, p = 0.044), having an overweight/obese BMI (AOR = 1.59, 95% CI: 1.03–2.46, p = 0.036), an irregular snack pattern (AOR = 1.73, 95% CI: 1.11–2.69, p = 0.014), moderate GSA (AOR = 2.41, 95% CI: 1.36–4.24, p = 0.002), and anxiety (AOR = 1.73, 95% CI: 1.14–2.64, p = 0.010). In contrast, individuals with a positive ARFID screen (AOR = 0.49, 95% CI: 0.26–0.93, p = 0.030) or functional dyspepsia (AOR = 0.47, 95% CI: 0.29–0.77, p = 0.003) had decreased likelihood of answering this question with a “yes.”

4. Discussion

We examined factors associated with a positive SCOFF eating disorder screening outcome among patients with hEDS/HSD and DGBI. The findings revealed that high GSA and NIAS' measured “fear of aversive consequences from eating” strongly predicted a positive SCOFF screening outcome, suggesting that worries around GI symptoms may drive positive responses on the SCOFF by mimicking eating disorder behaviors. Other predictors included an overweight/obese BMI, having an irregular snack pattern, anxiety, and depression. A visual summary of key findings discussion points is shown in Figure 1.

FIGURE 1.

Pathways influencing a positive SCOFF screen in hypermobile Ehlers–Danlos syndrome/hypermobility spectrum with comorbid disorders of gut–brain interaction. The diagram illustrates that standard SCOFF screening results (pink and red boxes) for this population could be driven by behaviors related to gastrointestinal symptoms (blue boxes), rather than eating disorders traits. SCOFF, (Sick, Control, One Stone, Fat, Food) eating disorder screening questionnaire.

4.1. Worries and Fear Around GI Symptoms Predict a Positive SCOFF Outcome in hEDS/HSD

High GSA strongly predicted a positive SCOFF outcome (Table 5). Item‐level analysis suggests that this association is driven by “yes” answers to the second and fifth questions (“Do you worry you have lost Control over how much you eat?” and “Would you say that Food dominates your life?”) (Table 6). These findings offer a novel explanation for observations reported in prior literature. Mari et al. [57], and Sultan et al. [58] found higher rates of positive SCOFF screens among patients with IBS with greater adherence to a low FODMAP elimination diet with the second or fifth being the most answered “yes.” Our study provides a novel finding by showing that in a logistic regression model accounting for various factors, high GSA might explain these findings, particularly, since individuals with high GSA are significantly more likely to engage in dietary restrictions [27].

Moreover, high scores on the NIAS “fear of aversive consequences” subscale, which measures fears from negative consequences of eating [54], increased the odds of a positive SCOFF screen (Table 5), mainly through SCOFF third question which asks about recent weight loss (Table 6). This is consistent with earlier work that discusses how oral or gastrointestinal manifestations of hEDS/HSD can foster fear of oral or abdominal pain and lead to clinically significant weight loss [42, 59].

These findings combined indicates that worries and fear around symptoms caused by eating consistently predicted a positive SCOFF outcome in hEDS/HSD regardless of the measurement scale used (VSI or NIAS), which may complicate the interpretation of positive screening results (Table 2). However, it is important to acknowledge that GSA has been documented in individuals with confirmed eating disorders [60], and therefore these findings do not preclude the possibility that positive SCOFF screens in our sample reflect genuine eating disorder risk.

4.2. Anxiety Predicts a Positive SCOFF Outcome in hEDS/HSD

The second strongest predictor of a positive SCOFF screen was anxiety with participants classified as having anxiety by the HADS being twice as likely to screen positive (Table 5). This is consistent with previous evidence showing that anxiety disorders are among the most common psychiatric commodities in individuals with eating disorders [61], and higher anxiety scores have been reported in individuals with positive SCOFF screens [62, 63].

4.3. Functional Dyspepsia Did Not Predict a Positive SCOFF Outcome in hEDS/HSD

Interestingly, none of the analyzed DGBI predicted a positive SCOFF outcome in the logistic regression model (Tables 5 and S1). On the contrary, individuals with functional dyspepsia were less likely to screen positive or respond with “yes” to the second and fifth SCOFF questions (“Do you worry you have lost control over how much you eat?” and “Would you say that food dominates your life?”) (Table 6). This could be due to the closer association of functional dyspepsia with ARFID symptoms [55, 64], an eating disorder not assessed by the SCOFF questionnaire.

4.4. ARFID Reduces the Probability of Positive SCOFF Outcome in hEDS/HSD

NIAS' “fear of aversive consequences” subscale significantly predicted SCOFF positive outcome, whereas “picky eating” and “lack of interest in food” did not. Furthermore, participants with a positive ARFID screen were less likely to screen positive on the SCOFF (Table 5). Literature have noted NIAS inability to distinguish between ARFID and other eating disorders [65, 66], however, the pattern shown in our findings suggest that questionnaire‐based eating disorder screening tools may accidentally capture eating behaviors stemming from underlying gastrointestinal symptoms. As described in Mikhael‐Moussa et al. recent review, ARFID presentations are common among those with DBGI, and their most frequently endorsed NIAS subscale is the “fear of aversive consequences” [66]. This have prompted researchers to raise concerns about the possibility of over‐diagnosing of ARFID in GI populations [67, 68], with some suggestion applying severity‐based criteria for ARFID to help draw a clearer distinction between ARFID and DGBI and reduce the diagnostic uncertainty in this population [69]. Therefore, in DGBI population such as patients with hEDS/HSD, we suggest that NIAS results should be interpreted cautiously and not be used in isolation for diagnostic purposes.

4.5. Overweight or Obesity and Female Gender Predicts a Positive SCOFF Outcome in hEDS/HSD

Participants with an overweight/obese BMI were significantly more likely to screen positive on the SCOFF, contrasting with previous reports of lower SCOFF sensitivity in populations with high obesity prevalence (Table 5) [47]. Finally, female participants were more likely to respond with “yes” to the fifth SCOFF question. However, the observed wide confidence interval indicates uncertainty in the effect estimate.

4.6. Self‐Reported AN or BN Did Not Predict a Positive SCOFF Outcome in hEDS/HSD

Although self‐reported history of AN or BN was associated with a positive SCOFF screen in the univariate analysis (Table 4), neither condition remained independently associated with a positive SCOFF screening outcome in the multivariable model (Table 5). However, self‐reported history of AN or BN were significant predictors of “yes” responses to SCOFF questions one and four (Table 6), which focus on induced vomiting and the distorted body image, behaviors highly specific to eating disorders. Given that SCOFF has demonstrated great accuracy in detecting AN or BN in young women, yet shows lower sensitivity in other populations [47], these results suggest the need for validation studies in a population of patients with hEDS/HSD.

4.7. Strengths and Limitations

Strengths include being the first study to explore each SCOFF item individually, offering novel insight into in individuals with hEDS/HSD. Additionally, the use of logistic regression is a key methodological advantage, as it is well‐suited for predicting outcomes while adjusting for confounding factors. Some limitations included in the study are that the comorbidities were based on self‐reports and were not confirmed through clinical assessment, which may affect the accuracy and reliability of the diagnostic data. Furthermore, given that nearly all individuals (99%) within our sample have at least one DGBI, the findings cannot be generalized to patients with hEDS/HSD without a DGBI.

4.8. Implication for Clinical Practice

Approximately 25% of our sample of hEDS/HSD/DGBI cohort have screened positive for eating disorders risk using the SCOFF questionnaire (Table 3), a rate similar to that observed in previous DGBI studies [57, 63, 70, 71]. Given such prevalence and the routine use of dietary modification to manage DGBI symptoms [72, 73], clinical guidelines [10] and authors [11, 13] have recommended using SCOFF as a screening tool before initiating DGBI treatment. However, concerns remain regarding SCOFF accuracy outside of its original context of AN and BN [47], with evidence suggesting it is more than often used incorrectly [74].

Our findings indicate that in hEDS/HSD with comorbid DGBI, a positive SCOFF screen holds dual interpretation where it can reflect risk of eating disorder or arise from alternative pathways (Figure 1), such as worries and fear of aversive consequences from eating. These factors primarily influenced responses to items two, three, and five (Table 6 and Figure 1). The phrasing in these three items may elicit misleading “yes” responses in populations where gastrointestinal symptoms influence eating behaviors or appetite. Future research should explore whether adaptations of SCOFF or other eating disorder screening tools for gastrointestinal populations could improve discrimination between eating disorder traits and behaviors related to gastrointestinal symptoms.

5. Conclusion

The applicability of SCOFF questionnaire in assessing true risk of eating disorders in patients with hEDS/HSD with comorbid DGBI remains inconclusive, particularly since SCOFF has not undergone psychometric validation within DGBI content [7]. However, even if the SCOFF does not reliably detect eating disorders risk in hEDS/HSD, a positive result may nonetheless signal significant anxiety and high GSA, thereby warranting further clinical evaluation and targeted intervention of these factors rather than exclusive focus on examination of eating disorders.

Author Contributions

L.D.M. conceptualized the study. L.D.M. and A.F. designed the study. R.A. performed the statistical analyses and drafted the initial manuscript. L.D.M. cowrote major sections of the manuscript and edited revisions. A.F. designed the statistical methodology and assisted with the interpretation of data. N.Z.‐L., C.V., R.T., and Q.A. reviewed and commented on the manuscript. All authors approved the final version.

Funding

Funding for the original CANDI‐hEDS study was provided by the Ehlers–Danlos Support UK, the Sandhu Charitable Foundation, and the Ehlers–Danlos Society. No further external funding was received for the conduct of the present study.

Ethics Statement

The original CANDI‐hEDS study protocol involved human participants and was approved by the Westminster Research Ethics Committee, London (REC reference 21/PR/0916).

Consent

All participants provided written informed consent to participate.

Conflicts of Interest

R.T., Q.A., and A.F. received the original grant by the Ehlers‐Danlos Support UK. The other authors declare no conflicts of interest.

Supporting information

Figure S1: Flow chart of data preparation and handling during the statistical analysis.

Table S1: Full results of the binary logistic regression analysis for predicting the probability of having a positive SCOFF eating disorder screening outcome.

Table S2: Full results of the binary logistic regression analysis for predicting the probability of a positive answer to SCOFF's first question “Do you make yourself Sick because you feel uncomfortably full?.”

Table S3: Full results of the binary logistic regression analysis for predicting the probability of a positive answer to SCOFF's second question “Do you worry you have lost Control over how much you eat?.”

Table S4: Full results of the binary logistic regression analysis for predicting the probability of a positive answer to SCOFF's third question “Have you recently lost more than One stone (6.4 kg) in a 3‐month period.”

Table S5: Full results of the binary logistic regression analysis for predicting the probability of a positive answer to SCOFF's fourth question “Do you believe yourself to be Fat when others say you are too thin?.”

Table S6: Full results of the binary logistic regression analysis for predicting the probability of a positive answer to SCOFF's fifth question “Would you say that Food dominates your life?.”

Data Availability Statement

The data used in this study were obtained from the CANDI‐hEDS study and are available on request at q.aziz@qmul.ac.uk. The data are not publicly available due to privacy and ethical restrictions.