Maladaptive weight control and eating behaviours in female adolescents/young adults are associated with increased risk of irritable bowel syndrome in adulthood: Results from the Growing Up Today Study (GUTS)

Summary

Background:

Irritable bowel syndrome (IBS) is common among individuals with eating disorders. The relationship between these conditions is probably bidirectional. However, data on the risk of IBS among those with prior eating disorders is largely limited to cross-sectional studies.

Aim:

To evaluate prospectively the association between maladaptive weight control/eating behaviours in females during adolescence/young adulthood with subsequent IBS using the Growing Up Today Study (GUTS)

Methods:

Starting in 1996 (Age: 9-14) and during follow-up, participants reported frequency of maladaptive eating/weight control behaviours during the past year to lose weight, self-induced vomiting (n=5740), laxative use (n=5438), and fasting (n=5522) in addition to reporting binge eating (n=4459). Starting in 2001 and during follow-up, participants reported if they had ever been diagnosed with an eating disorder (n=5316). Incident IBS cases were identified from four questionnaire cycles (2013, 2014, 2016, 2019), with participants specifying the year of diagnosis if occurring before the questionnaire date. Multivariable logistic regressions adjusting for age, body mass index, and depressive symptoms estimated the associations of interest.

Results:

Maladaptive weight control/eating behaviours were associated with increased IBS risk [ORs (95% CIs) for laxatives to lose weight = 3.67(2.52-5.35), vomiting to lose weight = 1.83(1.29-2.60), fasting to lose weight = 2.62(1.86-3.70), and bingeing = 2.25(1.54-3.28)] as was history of eating disorder diagnosis [OR (95% CI) = 3.42(2.38-4.90)]. Magnitude of IBS risk increased with the frequency of maladaptive behaviours .

Conclusions:

There is evidence for the potential role of early maladaptive weight control/eating behaviours in the development of adult IBS among females.

Graphical Abstract

Introduction

Irritable bowel syndrome (IBS), a disorder of gut-brain interaction, is one of the most common gastrointestinal conditions worldwide.¹ Emerging data has suggested the frequent coexistence of IBS and eating disorders in clinical practice.^2,3

Disordered eating behaviours are common among IBS patients and gastrointestinal symptoms may trigger maladaptive eating behaviours among a subset of patients using dietary modification for symptom relief. In a study including 100 IBS patients and 100 healthy adults, IBS patients reported significantly higher Eating Attitudes Test (EAT-26) score, an indicator of eating disorder risk.⁴ Conversely, in a cross-sectional study within the Swedish twin study, participants with behaviours of binge eating were more likely to meet criteria for IBS compared to participants without binge eating behaviour.⁵ In a survey of 234 participants with a history of or current diagnosis of an eating disorder, 64% met criteria for IBS.⁶ In another cross-sectional study of 955 IBS patients, 13% of patients endorsed severe food avoidance/restriction; these patients also reported lower quality of life, lower intake of nutrients, and more severe gastrointestinal symptoms compared to those without severe food avoidance/restriction.⁷

Indeed, recent evidence supports a bidirectional relationship between eating disorders and IBS.^2,3 However, prospective data on the association of earlier eating disorders and later IBS risk are very limited.⁸ Therefore, using prospective data from the Growing Up Today Study (GUTS), we aimed to examine if maladaptive weight control and eating behaviours in adolescent and young adult women are associated with risk of IBS in adulthood.

Methods

We conducted an analysis among females in the GUTS cohort, an ongoing prospective cohort of 9,039 female offspring of Nurses’ Health Study II participants at enrolment (baseline (1996) age: 9-14)⁹. Starting in 1996 and during follow-up (1996, 1997, 1998, 1999, 2001, 2003, 2005, 2010, 2013), participants reported frequency of maladaptive eating/weight control behaviours during the past year to lose weight: self-induced vomiting, laxative use, and fasting in addition to reporting binge eating.⁹ Specifically, “Vomiting/laxative use/fasting to lose weight” was assessed by asking how often during the past year did participants make themselves throw up/take laxatives/fast to lose weight or to keep from gaining weight.” Response categories included “never,” “less than once/month,” “1-3 times/month,” “once/week,” “2-6 times/week,” or “every day”. Binge eating was assessed by two questions: (1) the frequency of overeating episodes (“How often during the past year did you eat so much food in a short period of time that you would be embarrassed if others saw you (binge eating or gorging)” with similar response categories; and (2) For participants who reported overeating, they were directed to a yes/no follow-up question: “Did you feel out of control, like you couldn’t stop eating even if you wanted to stop?” Those who reported both overeating and loss of control were classified as bingeing.

Participants were classified as “ever” if they indicated any frequency of the behaviours other than “never”. Patients were further grouped into “at least monthly” and “at least weekly” if they practiced these behaviours on at least a monthly or weekly basis. Starting in 2001 and during follow-up, participants reported if they had ever been diagnosed by a healthcare provider with an eating disorder (i.e., binge-eating disorder, bulimia nervosa, anorexia nervosa, other eating disorders); we used this information as a secondary exposure.

The IBS question information was initially posed in the 2013 questionnaire, when participants were asked to report if they have been diagnosed with IBS in any of the following time intervals: before 2006, 2006, 2008, 2009-2011 and 2012+. Subsequent IBS diagnoses were collected via the 2014, 2015, 2016, and 2019 questionnaire cycles. We excluded all IBS cases diagnosed prior to 2006 to ensure all IBS cases were incident cases and occurred after the exposures of interest (i.e., maladaptive behaviours or eating disorder diagnosis). We used behaviour data (the exposure) occurring anytime up to 1 year before the IBS diagnosis to minimize the potential for reverse causation. In a subset of 115 IBS cases with available medical records for analysis, an expert (KS) blinded to eating disorder status reviewed each patient’s medical records and determined the accuracy of the self-reported IBS diagnosis using Rome III criteria applied to the clinician’s notes while ensuring that there was no organic disease subsequently diagnosed that could account for their symptoms. Chart review confirmed 74 definite IBS cases and 37 probable cases (diagnostic accuracy for IBS diagnosis of 96.5%). Although this data was analysed after the introduction of the more stringent Rome IV IBS criteria in 2016, our chart review used the Rome III criteria to be consistent with what clinicians would be using to diagnose IBS during the majority of the follow-up period.

Our final analysis included participants with complete data on IBS status and maladaptive eating/weight control behaviours [self-induced vomiting (n=5740, 425 IBS cases), laxative use (n=5438, 123 IBS cases), fasting (n=5522, 206 IBS cases), binge eating (n=4459, 186 IBS cases)] or diagnosed eating disorders (n=5316, 127 IBS cases)]. Covariates including age, body mass index (BMI) as well as depressive symptoms at the time of most recently reported maladaptive behaviours were used and presented. Participants with missing data on exposure (i.e., maladaptive weight control and eating behaviours) and/or outcome (i.e., IBS status) were excluded from our analytic sample. Depressive symptoms were assessed with the McKnight Risk Factor Survey IV (1999, 2001, 2003) or the Centre for Epidemiologic Studies Depression Scale (2007, 2010, 2013); consistent with cutoffs previously used in our cohort,¹⁰ participants in the top quintile of survey scores by either metric were categorized as "high depressive symptoms". Multivariable logistic regression models were built to estimate odds ratios (ORs) and 95% confidence intervals (CI) for IBS risk.

The study protocol was approved by the Institutional Review Boards of the Brigham and Women's Hospital and the Harvard T.H. Chan School of Public Health. SAS 9.3 for UNIX (SAS Institute Inc; Cary, NC) was used for all data analyses.

Results

Characteristics of study participants by maladaptive weight control and eating behaviours are shown in Table 1. Participants who engaged in maladaptive weight control/eating behaviours had higher rates of depressive symptoms but similar age and BMI.

Table 1.

Characteristics of study participants by maladaptive weight control and eating behaviours

Variables	Vomiting to lose weight (N = 5740)		Laxative use to lose weight (N = 5438)		Fasting to lose weight (N = 5522)		Binge eating (N = 4459)
	Never	Ever	Never	Ever	Never	Ever	Never	Ever
No. of participants	4823	917	4939	499	3677	1845	2610	1849
Age	27.5 (2.4)	27.6 (2.4)	27.9 (1.7)	27.8 (5.5)	27.8 (1.7)	27.7 (2.4)	27.9 (1.7)	27.6 (2.4)
BMI	24.6 (5.4)	25.1 (5.7)	24.7 (5.5)	24.8 (5.5)	24.3 (5.2)	25.6 (6.0)	24.3 (5.0)	26.0 (6.4)
High depressive symptoms	42.4	64.7	44.9	63.3	38.8	62.6	39.1	63.6

Notes: Abbreviation: No.=number. BMI=body mass index. Covariates represent those at the time of most recently reported maladaptive behaviour. Age and BMI are reported as mean with standard deviation (SD). BMI was calculated as weight(kg)/height²(m²) based on self-reported weight and height. Depressive symptoms were assessed with the McKnight Risk Factor Survey IV (1999, 2001, 2003) or the Center for Epidemiologic Studies Depression Scale (2007, 2010, 2013), consistent with cutoffs previously used in this cohort; participants in the top quintile of survey scores by either metric were categorized as "high depressive symptoms" and shown as %.

Compared with those who never engaged in the behaviour, each maladaptive behaviour was associated with an increased risk of IBS: vomiting for weight loss [ever: OR = 1.83 (95% CI 1.29 – 2.60), at least monthly: OR = 2.26 (95% CI 1.45 – 3.51), at least weekly: OR = 2.41 (95% CI 1.37 - 4.24)], laxative use for weight loss [ever: OR = 3.67 (95% CI 2.52 - 5.35), at least monthly: OR = 4.50 (95% CI 2.83 - 7.16), at least weekly: OR = 4.76 (95% CI 2.85 - 7.96)], fasting for weight loss [ever: OR = 2.62 (95% CI 1.86 - 3.70), at least monthly: OR = 3.14 (95% CI 2.13 - 4.63), at least weekly: OR = 3.18 (95% CI 1.94 - 5.19)], and binge eating [ever: OR = 2.25 (95% CI 1.54 - 3.28), at least monthly: OR = 2.65 (95% CI 1.74 - 4.03), at least weekly: OR = 2.87 (95% CI 1.72 - 4.79)] with laxative use for weight loss demonstrating the greatest risk elevation. More frequent maladaptive behaviours (i.e., monthly, weekly) were associated with an increased magnitude of IBS risk (Table 2).

Table 2.

Associations of maladaptive weight control and eating behaviours with IBS risk

		Never	Ever	At least monthly	At least weekly
Vomiting to lose weight	No of IBS cases	344	81	51	34
	No of Participants	4823	917	508	298
	Model 1 OR (95% CI)	ref	1.81 (1.28, 2.56)	2.25 (1.46, 3.48)	2.42 (1.39, 4.21)
	Model 2 OR (95% CI)	ref	1.82 (1.29, 2.57)	2.26 (1.46, 3.49)	2.42 (1.39, 4.20)
	Model 3 OR (95% CI)	ref	1.83 (1.29, 2.60)	2.26 (1.45, 3.51)	2.41 (1.37, 4.24)
Laxative use to lose weight	No of IBS cases	86	37	22	14
	No of Participants	4939	499	245	136
	Model 1 OR (95% CI)	ref	3.84 (2.67, 5.52)	4.78 (3.06, 7.48)	5.15 (3.12, 8.52)
	Model 2 OR (95% CI)	ref	3.84 (2.67, 5.52)	4.79 (3.06, 7.50)	5.18 (3.13, 8.56)
	Model 3 OR (95% CI)	ref	3.67 (2.52, 5.35)	4.50 (2.83, 7.16)	4.76 (2.85, 7.96)
Fasting to lose weight	No of IBS cases	63	143	81	36
	No of Participants	3677	1845	906	451
	Model 1 OR (95% CI)	ref	2.65 (1.92, 3.66)	3.19 (2.21, 4.59)	3.29 (2.07, 5.22)
	Model 2 OR (95% CI)	ref	2.66 (1.93, 3.68)	3.20 (2.22, 4.62)	3.31 (2.08, 5.24)
	Model 3 OR (95% CI)	ref	2.62 (1.86, 3.70)	3.14 (2.13, 4.63)	3.18 (1.94, 5.19)
Binge eating	No of IBS cases	46	140	88	40
	No of Participants	2610	1849	1026	501
	Model 1 OR (95% CI)	ref	2.20 (1.53, 3.17)	2.56 (1.71, 3.84)	2.78 (1.69, 4.56)
	Model 2 OR (95% CI)	ref	2.24 (1.55, 3.24)	2.60 (1.73, 3.91)	2.83 (1.72, 4.66)
	Model 3 OR (95% CI)	ref	2.25 (1.54, 3.28)	2.65 (1.74, 4.03)	2.87 (1.72, 4.79)

Notes: Abbreviation: No.=number. IBS=irritable bowel syndrome. OR=Odds Ratio. CI=Confidence Interval. Model adjustment: Model 1 is adjusted for age (continuous); Model 2 is adjusted for age and BMI (continuous); Model 3 is adjusted for age (continuous), BMI, and presence of high depressive symptoms (yes/no). Covariates (age, BMI, high depressive symptoms) represent those at the time of most recently reported maladaptive behaviour (vomiting to lose weight, laxative use to lose weight, fasting to lose weight, and binge eating).

In a secondary analysis, a history of a healthcare provider-diagnosed eating disorder (37 IBS cases among 403 participants) versus those without (90 IBS cases among 4913 participants) was also associated with increased IBS risk [OR=3.42 (95% CI 2.38, 4.90)].

Discussion

In a prospective cohort of female adolescents and young adults, we observed a two to more than three times increased risk of subsequent IBS among participants engaging in any earlier maladaptive weight control/eating behaviours. We also found a greater than three times greater odds of IBS in adulthood among adolescents/young adults with a prior diagnosis of eating disorder than those without eating disorders. These associations exhibited a “dose response” pattern, i.e., more frequent behaviours were associated with further increased risk of IBS.

Most previous research studying the intersection between eating disorders and disorders of gut-brain interaction such as IBS used small inpatient or outpatient samples with cross-sectional or retrospective designs.^3,11 Many previous studies focused on specific gastrointestinal symptoms, finding that functional gastrointestinal symptoms such as nausea, abdominal pain and bloating are common in patients with eating disorders.¹² Previous research also showed that functional dyspepsia (particularly the post-prandial distress subtype)¹³ and gastrointestinal dysmotility including oesophageal dysmotility, gastroesophageal reflux disease and symptoms,¹⁴ dysphagia,¹⁵ delayed gastric emptying,¹⁶ and delayed colonic transit ¹⁷ are common in individuals with eating disorders as well.^3,18

Co-occurrence of eating disorders and IBS has been demonstrated in many studies though most data were not able to clarify the temporal relationship due to study design.¹¹ Using National Health Insurance Research Database of Taiwan, Yeh et al. found a 7-fold increased risk of eating disorders in an IBS cohort as compared to a non-IBS cohort.¹⁹ In contrast (i.e., IBS risk in eating disorder patients), Perkins SJ et al. performed a small-scale survey in 234 participants with active or past eating disorders in a UK study and found 64% of these individuals had IBS; in fact, 87% of these patients reported having eating disorders on average 10 years before the development of IBS.⁶ In a case-control study of 765 eating disorder cases and 1240 controls, 88.2-95.5% of individuals with eating disorders reported at least one disorder of gut-brain interaction and 34.8-48.7% reported at least three disorders of gut-brain interaction. IBS was the most frequently reported disorder of gut-brain interaction (43.9-58.8%) among all eating disorder cases.²⁰ In a retrospective case-control study, binge eating and purging were associated with 2-fold and 3-fold elevated risk of all disorders of gut-brain interaction combined, respectively, including IBS, functional bloating, functional constipation, and functional diarrhoea among other unspecified functional bowel disorders. Of note, their results for risk of disorders of gut-brain interaction with previous laxative misuse and fasting were only marginally significant, which was likely due to a small sample size.²⁰ Our prospective data are consistent with these previous results and provide further evidence for the potential directionality of the association between disordered eating and IBS.

Although epidemiological data can only imply association, the strong magnitude and dose-response pattern of the associations between maladaptive eating behaviours and incident IBS as well as consistent findings across different behaviours do suggest a potential causative relationship. That said, eating disorders and disorders of gut-brain interaction, including IBS, share psychological risk factors including depression, anxiety, posttraumatic stress disorder, and childhood adverse events;^{21-23, 24, 25, 26, 27} thus, the development of eating disorders and IBS could be the result of shared, underlying confounders rather than a cause-and-effect relationship.

Mechanistically, excessive weight loss may exert detrimental effects on the function of the GI tract via impairment of nutrient absorption and energy balance, with excessive caloric restriction associated with maladaptive gut microbiome restructuring.²⁸ Laxative use has been linked to changed gut microbiome profiles as well.²⁹ Significant alterations in gut microbiota have been observed in patients with anorexia nervosa, presumably as a result of highly restricted diets.³⁰ These alterations in the gut microbiome may attenuate with refeeding, but an altered microbiome persists nonetheless, and gut microbiota composition continues to differ from healthy controls.^30-34 Such persistent changes in the gut microbiome could increase susceptibility to IBS and other gastrointestinal symptoms. In fact, a study of a cohort of 99 inpatients with eating disorders found that the majority (97%) met criteria for at least one disorder of gut–brain interaction on admission, with the rate falling to 77% over time.³⁵ However, 34% of these patients acquired symptoms of at least one new disorder of gut-brain interaction over subsequent follow up after discharge.

Although our secondary analysis showed an increased odds of IBS among those with diagnosed eating disorders, our primary analysis showed increased odds of IBS in disordered eating or maladaptive weight control behaviours even in those that likely did not meet criteria for a full-threshold DSM-5 eating disorder. Such findings suggest that even patients with “subthreshold” eating disorder symptoms (i.e., not meeting full criteria for any of the named eating disorders) may still be at increased risk for IBS and this could be informative for preventative interventions.

Our study has several advantages. First, to our knowledge, ours is the first prospective data linking maladaptive weight control/eating behaviours to IBS risk, with most previous evidence on this topic from cross-sectional data with high heterogenitiy.^8,12 Our exposures were collected via validated questionnaires and the questions themselves have precluded the possibility of IBS symptoms driving the examined maladaptive behaviours, because we specifically asked if participants adopted these behaviours to lose weight. Furthermore, medical record review confirmed the accuracy of the IBS diagnosis in a subset of patients where we had access to medical records. Nevertheless, we do acknowledge several limitations. First, we cannot exclude the possibility of residual confounding which is inevitable in any observational study. Second, although we asked participants to indicate year of their IBS diagnosis, we cannot exclude recall bias in the reporting of this information, and they may have had IBS symptoms prior to diagnosis as well. Also, we examined all IBS cases together as information for IBS subtypes (i.e., IBS with diarrhoea, constipation, or mixed) were not available in our cohort. We also examined all eating disorder cases together in our secondary analysis because the power to examine specific eating disorder diagnoses was limited. We also acknowledge that the same set of questionnaires was mailed to all participants of the GUTS cohort regardless of age; we cannot exclude the possibility that in the early phase of our cohort (baseline year: 1996), children of different ages (baseline age: 9-14) may understand certain questions such as “laxative use” differently (i.e., leading to misclassification bias), and we cannot be sure whether the participant or their parent completed the questionnaires at that time—though we do anticipate these differences would be minimized during longitudinal follow-up. Finally, all our participants were female, which may limit generalizability. However, both eating disorders and IBS are more prevalent in females,^3,21 so our study included a targeted sample at increased risk.

In conclusion, our data suggest that maladaptive weight control and eating behaviours in adolescence and young adulthood are associated with an elevated risk of IBS in adulthood. Recently, the US Preventive Services Task Force determined that there is insufficient evidence to support eating disorder screening with a normal or elevated BMI.³⁶ Our findings provide timely epidemiological evidence for the potential role of early maladaptive weight control/eating behaviours in the development of IBS in adulthood. Eating disorder screening and behaviour modifications among target populations may be an effective strategy for IBS prevention. Additional prospective studies are needed to replicate our findings and to understand the mechanisms through which early maladaptive relationships to food and body image put some females at risk for later IBS.

Data availability statement:

The data of this study are available upon reasonable request. Further information including the procedures to obtain and access data from the Growing Up Today Study is described at https://gutsweb.org/collaborate-with-guts/ (contact email: guts@channing.harvard.edu)