Psychiatric comorbidities are highly prevalent in non-esophageal eosinophilic gastrointestinal diseases

Abstract

Background and aims:

The prevalence of psychiatric comorbidity in non-esophageal eosinophilic gastrointestinal disorders (EGIDs) has not been studied. We aimed to ascertain the prevalence of psychiatric diagnoses and psychiatric medication use in children, adolescents, and adults with EGIDs and to assess whether psychiatric comorbidity affects clinical presentation.

Methods:

This was a retrospective cohort study of newly diagnosed patients with a non-esophageal EGID at the University of North Carolina from 2008 – 2020. Psychiatric diagnoses and medications were extracted from medical records. We compared the clinical and demographic features of EGID patients with and without psychiatric diagnoses.

Results:

Of 79 patients (mean age 23.3, 53% male, 78% white) with a non-esophageal EGID diagnosis, 40 (51%) were diagnosed with a co-morbid psychiatric disease. Anxiety (37%) and depression (28%) were most common. There were also 40 (51%) patients treated medically for a psychiatric diagnosis. Patients with a psychiatric diagnosis were more commonly ≥18 years of age at time of EGID diagnosis (OR 3.95, 95% CI: 1.20 – 13.02), and to have endorsed symptoms of nausea (OR 5.31, 95% CI: 1.33 – 21.22) and dysphagia (OR 4.24, 95% CI: 1.18 – 15.26).

Conclusion:

Psychiatric diagnoses were very common in non-esophageal EGID patients with approximately 7 in 10 adults and one-third of children diagnosed. Similar proportions were found for psychiatric medication use. We also found that psychiatric illness may influence age of clinical presentation and symptoms. Providers should assess for concomitant psychiatric comorbidities in EGID patients.

Introduction

Eosinophilic gastritis, enteritis, and colitis, the non-esophageal eosinophilic gastrointestinal diseases (EGIDs), are chronic immune-mediated disorders characterized by gastrointestinal dysfunction and eosinophilic-predominant tissue inflammation.^1–4 The pathophysiology of EGIDs remains incompletely described. The condition likely derives from an aberrant Th2 immune response to food allergens,^5–7 although new data suggest mast cells may play a role as effector cells as well.⁸ EGIDs present with troublesome symptoms including abdominal discomfort, diarrhea, nausea and/or vomiting, bloating, and when severe, may manifest with anemia, lower extremity edema through protein-losing enteropathy, or eosinophilic ascites.^3,4,9–13 Symptoms specific to pediatric patients include growth retardation, pubertal delay, and failure to gain weight.¹⁴

Though EGIDs deleteriously impact patients’ clinical well being, these disorders also negatively affect patients’ health-related qualify of life (HRQOL), as EGID patients experience delayed and often missed diagnoses, and are faced with few effective treatment options.^10,15–18 Though prior studies assessed the impact of an EGID on HRQOL,^15,16 the impact of EGID specifically on patients’ mental health, which is distinct from HRQOL, is not well-described.¹⁹ Moreover, this is an unmet need, as patients describe psychosocial issues as important to their care.¹⁵ Given this dearth of information, the prevalence and types of psychosocial diagnoses and centrally acting medication use have not been described in EGID patients. Additionally, no study has evaluated whether the presence of comorbid psychiatric disorders or psychiatric medication use influences the clinical, endoscopic, or histologic features of EGID.

The aim of this study was to assess the prevalence and spectrum of psychiatric comorbidities and centrally acting medication use among a cohort of patients of all ages with an EGID diagnosis. The study also evaluated whether a diagnosis of a comorbid psychiatric disease was associated with the clinical, endoscopic, and/or histologic features of EGID at the time of an EGID diagnosis.

Methods

We conducted a retrospective cohort study at the University of North Carolina (UNC) from 2008–2020 utilizing the UNC EGID Clinicopathologic Database.^9,11 The UNC IRB approved this study. The database, which has been previously described,^9,11 contains all patients with a formal diagnosis of an EGID at UNC. Included patients demonstrated ≥20 peak eosinophils/hpf (eos/hpf) on gastric, ≥ 30 eos/hpf on duodenal, and/or ≥ 50 eos/hpf on colonic biopsy, gastrointestinal symptoms, no known secondary cause of gastrointestinal eosinophilia, and chronic GI symptoms. No consensus guidelines currently exist to define diagnostic histologic thresholds for increased eosinophilic mucosal infiltrates of the stomach, small bowel, or colon to diagnose an EGID. However, expert recommendation has suggested 30 eos/hpf in 5 hpf’s to define eosinophilic gastritis.²⁰ Prior work from our group, however, has demonstrated that less stringent and lower cell counts, as utilized in this paper, maintain excellent specificity for the diagnosis of an EGID,⁹ so we have elected to retain all subjects in this study with the diagnosis at our institution. As noted below, peak eosinophil count thresholds at the time of diagnosis were well above both potential thresholds.

We searched the electronic medical record of each patient within the UNC EGID Clinicopathologic Database to identify psychiatric comorbidities. The psychiatric comorbidities captured for this study included anxiety, depression, post-traumatic stress disorder (PTSD), bipolar depression, attention deficit disorder (ADD), and schizophrenia. Each psychiatric diagnosis was considered present if diagnosed by a provider in the medical record. Additionally, for each patient within the database, we extracted whether a prescription of a psychiatric medication was documented (e.g. selective serotonin reuptake inhibitor [SSRI], serotonin norepinephrine reuptake inhibitor [SNRI], tricyclic antidepressant, benzodiazepine, neuroleptic, buspirone, amphetamine, and other [i.e. pregabalin, gabapentin, trazodone, mirtazapine). We also recorded whether counseling specifically for a psychiatric condition had been performed. For medications, we ensured that the prescription was not for an incidental or otherwise explainable indication (e.g. a benzodiazepine for a procedure, gabapentin for a seizure disorder, pregabalin for diabetic peripheral neuropathy were not counted specifically toward the treatment of a psychiatric disorder).

Within a standardized data collection form, we extracted cohort demographics, symptoms, endoscopic findings, and histologic findings from the electronic medical records for all patients at the time of EGID diagnosis.

For analysis, descriptive statistics characterized the patient cohort. Bivariate statistics with Pearson’s chi-square test compared categorical variables. Student’s t-test compared bivariate continuous variables. We used multivariate regression models to assess the relationship between the clinical features of EGID and the presence of a psychiatric diagnosis. As described below, an individual model was created to assess each exposure by outcome relationship (e.g. an individual model was run for dysphagia, an individual model was run for nausea, an individual model was run for gastric erosions, and so forth). This modeling strategy was chosen, as there were limited outcomes (i.e. patients with psychiatric disease), thus resulting in over-fitting and spurious results if all covariates of interest were placed in a singular model. To find potential confounding variables, we ascertained whether any covariates were unequally distributed between patients with and without a psychiatric comorbidity by percentage for categorical variables and means for continuous variables. Additionally, to evaluate for confounding, we assessed the distribution of potential confounding variables by exposure variable(s) of interest. We found that age at EGID diagnosis, sex, and race (modeled dichotomously as white versus non-white) consistently confounded the exposure variable by psychiatric diagnosis relationship, which were utilized as covariates in each model to control for confounding. We next performed bivariate analyses to assess for unadjusted relationships between psychiatric disease presence and the predictor variable of interest (e.g. gender, race, age at EGID diagnosis, atopic disease diagnosis, and baseline eosinophil count). Lastly, multivariate logistic regression was used to estimate the independent variable by outcome odds ratio for each model, adjusted for confounding. As an example, we ran a unique multivariate logistic regression model to assess whether dysphagia as the exposure variable independently predicted the odds of a psychiatric disease diagnosis after controlling for age at EGID diagnosis, sex, and race.

All statistical analyses were performed using Stata 14.2 (Stata Corp, College Station, TX). Additional non-parametric testing did not substantively change the conclusions of this study. Non-statistical analyses comparing our results with nationally representative data were also completed to compare the relative frequency of psychiatric comorbidities in the US population with that seen in our EGID cohort.^21–28

Results

Baseline characteristics of EGID cohort at time of diagnosis

Of 79 patients with a non-esophageal EGID, the mean age at EGID diagnosis was 23.3 years with 46 (58%) diagnosed prior to 18 years of age. Most patients were white (78%) and male (53%). Of the 79 patients with an EGID, there were 36 (46%) with concurrent esophageal eosinophilia. There were 52 (66%) with gastric eosinophilia, 35 (45%) with small bowel eosinophilia, 21 (48%) with colonic eosinophilia, and 50 (63%) with multi-site involvement. We found that 40 (51%) had been diagnosed with a co-morbid psychiatric disorder. Of these 40 patients, a psychiatrist made the diagnosis of a psychiatric disorder in 18 (45%), the primary care physician made the diagnosis in 10 (25%), a GI in 2 (5%); the remaining 10 diagnoses were made outside of our system.

Types of psychiatric comorbidities and prevalence of psychiatric medication use

For the 40 patients with a psychiatric diagnosis, the most common diagnosis was anxiety (37%) followed by depression (28%). Other psychiatric diagnoses were less frequently made, but included ADD (11%) and bipolar depression (3%); none had schizophrenia. Adults were more likely than children to have received a psychiatric diagnosis (70% vs. 37%; p = 0.004) (Table 1). With the exception of ADD (6% vs. 15%; p = 0.21), specific psychiatric diagnoses were also more common in adults than children, with anxiety (52% vs. 26%; p = 0.02) and depression (49% vs. 13%; p = 0.001) again being the most prevalent conditions.

Table 1.

Baseline demographics factors, symptoms, endoscopic findings, and eosinophil counts in eosinophilic gastrointestinal disease patients with or without a concomitant psychiatric diagnosis

	Psychiatric diagnosis (n = 40)	No diagnosis (n = 39)	P value
Demographics
Age at diagnosis (mean ± SD)	26.8 ± 19.3	17.7 ± 22.3	0.06
Adult (≥ 18 years; n, %)	23 (58)	10 (26)	0.004
Symptom length (n, %)1	5.5 ± 10.8	4.1 ± 6.6	0.52
Male (n, %)	17 (43)	25 (64)	0.05
White (n, %)	29 (88)	22 (69)	0.06
Atopic diagnosis (n, %)	22 (55)	29 (74)	0.07
Symptoms (N, %)
Dysphagia	17 (45)	7 (19)	0.02
Food impaction	7 (18)	1 (3)	0.03
Heartburn	13 (36)	6 (19)	0.11
Chest pain	(19)	0 (0)	0.008
Abdominal pain	29 (83)	19 (59)	0.03
Nausea	23 (64)	(23)	0.001
Diarrhea	15 (38)	12 (31)	0.53
Endoscopy findings (N, %)
Normal EGD2	11 (28)	16 (41)	0.21
Esophageal eosinophilia	18 (49)	18 (50)	0.91
Normal esophagus	19 (48)	26 (67)	0.09
Esophageal rings	(25) 5	4 (10) 2	0.09
Esophageal stricture	(13)	(5)	0.25
Esophageal furrows	(28)	7 (18)	0.31
Esophageal plaques	6 (15)	6 (15)	0.96
Esophageal edema	4 (1)	3 (8)	0.72
Normal stomach3	11 (46)	14 (50)	0.77
Gastric erosion	3 (13)	1 (4)	0.23
Gastric erythema	8 (33)	9 (32)	0.93
Gastric nodularity	3 (13)	4 (14)	0.81
Gastric ulcer	4 (17)	5 (18)	0.91
Gastric stenosis	1 (4)	0 (0)	0.28
Normal duodenum4	14 (64)	11 (85)	0.18
Duodenal erosions	2 (9)	1 (8)	0.87
Duodenal erythema	2 (9)	1 (8)	0.89
Duodenal nodularity	1 (5)	0 (0)	0.44
Duodenal ulcer	3 (14)	0 (0)	0.16
Duodenal stenosis	4 (18)	0 (0)	0.10
Normal colon5	4 (36)	2 (20)	0.41
Colonic erythema	0 (0)	1 (10)	0.28
Colonic nodularity	2 (12)	0 (0)	0.34
Colonic stricture	1 (6)	0 (0)	0.51
Peak eosinophil count6
Gastric3	122.4 ± 123.8	66.1 ± 46.9	0.03
Small bowel4	65.2 ± 31.5	63.2 ± 29.3	0.85
Colon5	86.0 ± 17.0	99.9 ± 77.0	0.57

¹Symptom length: Length of symptomatic period preceding diagnosis ± standard deviation

²EGD: esophagogastroduodenoscopy

³Among patients with eosinophilic gastritis

⁴Among patients with eosinophilic enteritis

⁵Among patients with eosinophilic colitis

⁶Max peak eosinophil count: eosinophils per high-power field ± SD standard deviation

For the entire cohort, pharmacologic treatment with a psychiatric medication was common, with 40 (51%) being treated with a psychiatric medication. There was also documentation of 13% of the cohort undergoing counseling for a psychiatric diagnosis. As with psychiatric diagnoses, the utilization of a psychiatric medication was more common overall in adults compared with children (73% vs. 35%; p <0.0001). The most frequently prescribed medications for adults and children included SSRIs (33% vs. 13%; p = 0.03), SNRIs (24% vs. 7%; p = 0.03), and benzodiazepines (21% vs. 11%; p = 0.21). Among the 39 patients without a documented psychiatric diagnosis, there were 5 (13%) treated with a psychiatric medication. Indications for treatment with a psychiatric medication in this group included chronic nausea (3 patients) and chronic abdominal discomfort (1 patient). There was a single patient in whom the indication could not be discerned.

Among the 40 patients with a psychiatric diagnosis, the majority (65%) was diagnosed with their psychiatric comorbidity after the diagnosis of an EGID. We found that patients diagnosed with a psychiatric illness prior to their EGID diagnosis tended to have a longer duration of symptoms before receiving their EGID diagnosis when compared to those diagnosed with a psychiatric illness following EGID diagnosis (9.5 vs. 3.0 years; p = 0.09).

Clinical features of EGID patients with and without a psychiatric comorbidity

EGID patients with a psychiatric diagnosis were more likely to be ≥ 18 years old (58% vs. 26%; p = 0.004), female (57% vs. 36%; p = 0.05), and tended to be white (88% vs. 69%; p = 0.06) compared to those without a psychiatric comorbidity (Table 1). Length of symptoms prior to EGID diagnosis was comparable between these groups (5.5 vs. 4.1 years; p = 0.52). EGID patients with a psychiatric comorbidity more frequently reported dysphagia (43% vs. 18%; p = 0.02), chest pain (18% vs. 0%; p = 0.008), abdominal pain (73% vs. 49%; p = 0.03), and nausea (58% vs. 18%; p = 0.001). There was a numeric, though non-statistically significant difference in reported heartburn (33% vs. 16%; p = 0.11). The association between presence of a psychiatric disorder and dysphagia persisted after controlling for the presence of EoE in a multivariate model (OR 4.33; 95% CI 10.7 – 17.6). Endoscopic features did not significantly differ between those with and without a psychiatric diagnosis.

Age, dysphagia, and nausea were found to be independently associated with the presence of a psychiatric comorbidity after adjusting for race, age at diagnosis, and sex (Table 2), whereas endoscopic findings and peak eosinophil counts were not.

Table 2.

Unadjusted and adjusted odds ratios for associations between psychiatric diagnosis and demographics, symptoms, endoscopic, and histologic features

	Unadjusted OR (95% CI)	Adjusted OR (95% CI)
Demographics (N, %)
Adult1	3.92 (1.51 – 10.2)	3.95 (1.20 – 13.02)
Symptom length2	1.02 (0.96 – 1.08)	0.99 (0.92 – 1.07)
Male	0.41 (0.17 – 1.02)	0.66 (0.22 – 1.95)
White	3.29 (0.91 – 11.91)	2.08 (0.52 – 8.27)
Atopic diagnosis	0.42 (0.16 – 1.09)	0.29 (0.08 – 1.02)
Symptoms (N, %)
Dysphagia	3.35 (1.18 – 9.53)	4.24 (1.18 – 15.26)
Food impaction	8.09 (0.94 – 69.35)	5.12 (0.52 – 50.53)
Heartburn	2.45 (0.80 – 7.49)	2.37 (0.57 – 9.84)
Chest pain	1.00	1.00
Abdominal pain	3.31 (1.07 – 10.21)	3.99 (0.97 – 16.32)
Nausea	5.81 (1.96 – 17.21)	5.31 (1.33 – 21.22)
Diarrhea	1.35 (0.53 – 3.43)	0.17 (0.03 – 0.95)
Endoscopy findings (N, %)
Normal EGD3	0.55 (0.21 – 1.40)	0.84 (0.25 – 2.84)
Normal esophagus	0.45 (0.18 – 1.12)	0.60 (0.19 – 1.87)
Normal stomach4	0.85 (0.28 – 2.52)	2.25 (0.45 – 11.26)
Normal duodenum5	0.32 (0.06 – 1.81)	0.30 (0.02 – 3.68)
Normal colon6	2.29 (0.32 – 16.51)	7.49 (0.43 – 129.08)

¹Adult: comparison of proportion of patients ≥ 18 years old at diagnosis and symptom length adjusted

²Symptom length: increased odds for psychiatric diagnosis per each year of age added

³EGD: esophagogastroduodenoscopy

⁴Among patients with eosinophilic gastritis

⁵Among patients with eosinophilic enteritis

⁶Among patients with eosinophilic colitis

⁷Max peak eosinophil count: comparison of predicted max eosinophil count

We also qualitatively compared the prevalence of psychiatric diagnoses and prescription of psychiatric medications in our EGID cohort with reported prevalences within the US. For our patients, 70% of adults with an EGID had a diagnosis of any psychiatric condition; this compared to 21% within the US adult population.²¹ We also found that depression (49% vs. 8% US population) and anxiety (52% vs. 19%) were more common for the adults within our cohort (Figure 1A).²¹ This was also seen for comparisons with centrally acting medications (Figure 1B). The receipt of any psychiatric medication was 73% in our adult patients compared with 11% for US adults.²² We also found that individual use of psychiatric medications was more common in our patients. There were 33% of EGID patients prescribed an SSRI, which compared with 13% of US population receiving any form of anti-depressant for psychiatric illness.²³

Figure 1.

Psychiatric illnesses and psychotropic medication use in the study eosinophilic gastrointestinal disease (EGID) population and in the general population in the United States.^21–28 (A) Proportion of patients with any psychiatric illness, depression, or anxiety. (B) Proportion of patients with any psychotropic medications, antidepressants, or anxiolytics.

When compared to national averages, the prevalence of any psychiatric diagnosis was more common among pediatric patients included in the study (Figure 1A). In this cohort, 37% of children were diagnosed with a psychiatric condition, which is more than double the prevalence seen nationally (17%).²⁴ Psychiatric disorders most commonly found in our pediatric cohort included depression (13% vs. 2% nationally), anxiety (26% vs. 0.4%), and ADD (15% vs. 12% nationally). Additionally, prevalence of psychiatric mediation prescriptions were more commonly seen among our pediatric cohort when compared to national averages (Figure 1B).^25–28

Discussion

Diagnosis of a non-esophageal EGID has a deleterious impact on patients’ clinical well-being through troublesome symptoms and decreases patients’ HRQOL through delayed diagnoses and lack of effective therapies.^10,15,16 Though patients with EoE have been recently shown to have an increased burden of psychiatric diseases, the association between an EGID and patients’ mental health has not been fully explored.^19,29 In this study, we aimed to determine the prevalence and types of psychiatric diagnoses and psychiatric medication use among EGID patients across the age spectrum. We also explored the association between the presenting clinical features of an EGID and the presence of a psychiatric comorbidity. We observed that psychiatric comorbidities and psychotropic medication use were strikingly common in both adults and children with an EGID. Approximately 7 in 10 adults and nearly one-third of children received a diagnosis; similar proportions were prescribed a psychiatric medication. We also found that EGID patients with a psychiatric diagnosis were more likely to be older and have different symptoms on presentation (dysphagia and/or nausea) when compared to those without a psychiatric diagnosis. Interestingly, the majority of patients with a psychiatric diagnosis were diagnosed with their psychiatric comorbidity after being found to have an EGID, and these patients had a longer symptomatic period preceding their EGID diagnosis. Compared with nationally representative numbers from the general United States population, most of the psychiatric diseases and medication classes evaluated in this study were far more commonly diagnosed and prescribed in patients with non-esophageal EGIDs. This was true for both adults as well as pediatric patients, and emphasizes that mental health disorders are a new and substantial burden to an EGID diagnosis.

The mechanism of the relationship between psychiatric disease and a concurrent EGID diagnosis is not known. However, this observed association likely is multifactorial. Potential etiologies for these associates include burdensome and expensive treatments,³⁰ as well as lack of awareness and social support surrounding the diagnoses.¹⁵ Moreover, pathologic gastrointestinal eosinophilic infiltrates may impair gastrointestinal neuronal structure and function, as has been reported in the association between functional GI disorders and potential GI inflammation and eosinophilia.³¹

The impact of an EGID on patients’ HRQOL has been more fully described in EoE patients rather than in patients with more distal EGIDs.^32–35 This was best illustrated in a 2018 systematic review of 22 studies that showed that EoE disrupts and interferes with the daily life of EoE patients.³² These authors also documented that the severity of EoE symptoms is associated with the impact on HRQOL by the disease, and that the impact on HRQOL may attenuate with treatment of EoE specifically.

Limited data exist regarding psychological dysfunction experienced by patients with an EGID. In one survey of adults and adult caregivers of children with an EGID, the presence of an EGID was found to produce barriers across the medical, healthcare, social, and emotional domains of the survey. For instance, 64% of respondents reported significant stress due to out-of-pocket EGID costs, and 50% felt that their medical provider was not knowledgeable regarding their disease.¹⁵ A qualitative study of seven patients also showed that EGID elicits a range of psychological impacts ranging from emotional distress to depression following EGID disease onset.¹⁶ In another study of pediatric patients with EoE or other EGIDs, the presence of depression was associated with non-adherence to therapy, but the overall prevalence of depression or treatments were not described.³⁶ This work, in addition to our current study, highlights the critical need for screening for depression in all EGID patients, as features of depression may go undetected or misinterpreted, ultimately affecting patients’ mental health and deleteriously affecting EGID management.

Limitations for this study include the possibility of measurement bias given the retrospective design. We attempted to minimize measurement bias by utilizing strict definitions of our included variables and confirming that psychotropic medications were specifically prescribed for a psychiatric condition. However, as psychiatric diagnoses and psychotropic medications were assessed for retrospectively, if present, this bias likely resulted in the under-recognition of these diagnoses and medications. Additionally, because we could not utilize validated symptom measures and/or standardized instruments in a prospective fashion, symptom reports and endoscopic findings should be interpreted with caution. Our sample size was limited to 79 patients, which may also limit generalizability.

There are multiple strengths to this paper as well. Per our understanding of the medical literature, this study represents the largest and only cohort study specifically assessing for the association between psychiatric disease and features of an EGID. It also represents the only study to assess the prevalence and types of psychiatric diseases in a cohort of EGID patients.

In conclusion, psychiatric comorbidities were very common in a large cohort of patients with an EGID. We found that approximately 7 in 10 adults and nearly one-third of children with an EGID had a comorbid psychiatric diagnosis and similar proportions had a psychiatric medication prescribed. After adjusting for confounding variables, EGID patients with a psychiatric diagnosis were more likely to have reported dysphagia, nausea, and to be older than 18 at the time of EGID diagnosis. As the prevalence of psychiatric comorbidities were much higher than that seen in the general US population, this suggests that psychiatric disease should be assessed for in all patients with a confirmed EGID diagnosis.^21,24 Additional studies should assess whether the presence of a psychiatric disease influences the development of an EGID, or whether psychiatric disease derives from the consequences of living with an EGID.^21–28

What You Need to Know.

• Background: Symptoms of non-esophageal eosinophilic gastrointestinal disease (EGID) impact patients’ health-related quality of life. However, the prevalence of psychiatric disease in non-esophageal EGIDs has not been studied.

• Findings: Psychiatric comorbidities and psychotropic medication use were very common in patients of all ages with non-esophageal EGIDs. There were approximately 7 in 10 adults and one-third of children diagnosed with a psychiatric comorbidity. Moreover, psychiatric illness influenced non-esophageal EGID presentation.

• Implications for patient care: Psychiatric disease should be assessed for in all patients with a confirmed non-esophageal EGID.

Abbreviations:

ADD

Attention deficit disorder

EGIDs

eosinophilic gastrointestinal diseases

eos/hpf

eosinophils per high-power field

HRQOL

health-related qualify of life

PTSD

post-traumatic stress disorder

SNRI

serotonin norepinephrine reuptake inhibitor

SSRI

selective serotonin reuptake inhibitor

UNC

University of North Carolina