Prevalence of oesophageal eosinophilia and eosinophilic oesophagitis: a prospective study in unselected patients presenting to endoscopy

Abstract

Oesophageal eosinophilia (EE) is encountered in clinical practice as oesophageal biopsies are being obtained in patients with GI symptoms other than classical symptoms of eosinophilic oesophagitis (EoE). The prevalence, determinants and clinical relevance of EE identified irrespective of symptoms are unclear.

Aim

To determine the prevalence and risk factors of EE with or without EoE in a non-selected group of patients undergoing endoscopy and in primary care patients.

Methods

A cross-sectional study in a single VA center in which we obtained at least one oesophageal biopsy from patients presenting to elective endoscopy, as well as a sample of patients eligible for screening colonoscopy recruited from primary care clinics. EE was defined by > 15 eosinophils in a single HPF; and EoE was defined as definite, probable or none depending on the presence of EE, acid-suppressive therapy and oesophageal symptoms.

Results

EE was identified in 33 of 1357 patients (2.4%, 95% CI: 1.7–3.4); of whom 9 had definite EoE (0.66%, 95% CI: 0.23–1.10), 17 had probable EoE (1.25%), and the only 7 patients had EE without EoE. The prevalence of EE was 2.3% among patients undergoing elective endoscopy and 0.1% among patients eligible for screening colonoscopy. Seasonal allergies (adjusted OR: 2.78; 95%CI 1.26 – 6.11) and oesophageal strictures (4.50; 0.90 – 22.40) were associated with EE.

Conclusions

The prevalence of EE was 2.3% among unselected patients presenting to endoscopy most of whom have EoE. EE was present in 0.1% in primary care patients none of whom had EoE.

Background

Eosinophilic oesophagitis (EoE) was first reported in 1978,¹ and it has gained increased recognition as an etiology of dysphagia and food impaction in adults². The hallmark of EoE is oesophageal eosinophilia (EE), commonly defined by at least 15 eosinophils per high-power field, combined with symptoms of oesophageal dysfunction.^2,3 Other etiologies of EE may include gastro-oesophageal reflux disease (GERD), eosinophilic gastrointestinal disease and inflammatory bowel disease.

The prevalence of EoE has been examined by our group in a recent systematic review of the literature published through 2009.¹¹ We identified 9 studies examining the prevalence of EE related to EoE in adult populations, most of which evaluated patients presenting with food bolus obstruction, dysphagia or GERD symptoms, ^{4,5,6,7,8,9,10}. The reported EoE prevalence estimates ranged between 0.9% and 48.2%.

Most patients with EE and dysphagia or food impaction have EoE. However, EE irrespective of EoE is encountered more frequently in clinical practice than it did previously, as more oesophageal biopsies are being obtained during the evaluation of upper gastrointestinal symptoms. The prevalence, determinants and clinical relevance of EE identified irrespective of EoE symptoms are unclear.

Two population-based studies have reported the prevalence of EE or EoE. One reported a prevalence of 0.02% for EoE, but not EE, in 100,000 individuals,¹² and the second reported a prevalence of 1.1% for EE only, but not EoE, in 1,000 individuals¹³. The only 2 clinic-based studies that prospectively examined patients presenting to endoscopy for all indications reported an EoE prevalence rate of 0.9% in 4,146 patients⁹ and 6.5% in 400 patients¹⁰; however, both studies defined EoE exclusively by the presence of EE.

The aim of this cross-sectional study was to prospectively determine the prevalence, risk factors and clinical relevance (signs and symptoms) of EE among consecutively enrolled eligible patients presenting to elective EGD, as well as a background population representing primary care patients. In both groups, we performed systematic recording of endoscopic findings, oesophageal biopsies, and determination of risk factors by means of a survey questionnaire.

Materials and Methods

Study Population

The cases and controls in this study were obtained from two cross-sectional study conducted among eligible patients who were scheduled for a non-urgent oesophagogastroduodenoscopy (EGD) at the Michael E. DeBakey Veterans Administration Medical Center in Houston, TX, and a group of patients eligible for screening colonoscopy who were recruited from the primary care clinics at the same hospital to undergo a study EGD in addition to the colonoscopy. We excluded patients with active malignancy, oesophageal surgery, cirrhosis, or ongoing anticoagulant use. Written informed consent was obtained from all patients prior to enrollment. The institutional review boards at both the Michael E. DeBakey VA Medical Center and Baylor College of Medicine approved this study.

Data Collection

Before the study EGD, each patient was asked to complete a questionnaire to identify the onset time for GERD symptoms and then measure symptom frequency and severity for the preceding year. Information on medications, including proton pump inhibitors (PPIs), histamine type 2 receptor antagonists, and non-steroidal anti-inflammatory drugs, was collected, using a structured questionnaire, and verified by medical and pharmacy records in the subset of patients with EE. Information on symptoms, personal history of asthma, seasonal allergies, lifetime tobacco and alcohol use was also obtained using pre-EGD structured questionnaires.

Study EGD

All participants underwent a study EGD. During the EGD, at least one standardized single oesophageal mucosal biopsy was obtained, using Jumbo biopsy forceps 2–3 cm above the normal squamocolumnar junction, as well as biopsies from suspected Barrett’s oesophagus (BE) and several gastric biopsies. Based on the histopathological interpretation of the biopsies by two pathologists, EE was defined by the presence of >15 eosinophils per a single, high-powered field in at least one oesophageal biopsy. Definite EoE was defined as the histological presence of EE combined with oesophageal dysfunction symptoms (dysphagia, heartburn, chest pain) and concurrent use of acid-suppressive medication. Probable EoE was defined as EE combined with either oesophageal symptoms or use of acid-suppressive medication. Endoscopically suspected BE was recorded based on the Prague CM classification¹⁴. Definitive BE was considered in the presence of intestinalized columnar epithelia (confirmed by alcian-PAS stain) in the histopathological examination of biopsy samples obtained from suspected BE areas. Endoscopic BE only was defined by the presence of suspected BE in the absence of specialized intestinal epithelium. Subjects were defined as positive for H. pylori if organisms were seen on histopathology of any of the study gastric biopsies, or if review of the medical record showed a previous positive biopsy, presence of serum antibodies, or treatment received. A manual review of pathology records of was conducted to determine the number of collected oseophageal biopsies.

Data Analysis

Patients with EE/EoE served as our cases, while those without these diagnoses served as our controls. We calculated the prevalence of EE and EoE as the proportions (and accompanying 95% confidence intervals [CIs]) of patients with these conditions stratified by recruitment source (endoscopy, primary care). Potential risk factors, symptoms, and endoscopic signs were compared between cases and controls. In the unadjusted analyses, chi square tests were used to calculate categorical variables and t-tests were used to calculate continuous variables. In three separate multivariable regression models, we examined associations between EE and potential risk factors, symptoms, or endoscopic findings while adjusting for age and race/ethnicity. The analyses were also stratified by recruitment source. The models were constructed using stepwise backward elimination using p>0.25. Parameter estimates and standard errors were used to calculate odds ratios and 95% CIs.

Results

A total of 1357 consecutive eligible patients underwent EGD and fulfilled the criteria for this analysis; 997 (73.5%) were from those initially referred to elective EGD, and 360 (26.5%) were from the group recruited from primary care clinics for a screening colonoscopy and a study EGD. Age of participants with EE in the endoscopy group was 61.5 (IQR: 55.3–66.8) (Supplemental Table 1) and for the two patients with EE in the colonoscopy group it was 65.0 (IQR: 59.2–72.7); these were not different from the underlying group of participants where the median age for the endoscopy group was 61.5 (IQR: 55.9–65.9) and for the colonoscopy group it was 61.9 (IQR: 37.4–66.2). Approximately 67.7% were White and 29.5% were Black (Supplemental Table 2). Approximately 11.7% (4/34) of cases and 7.0% of controls had at least two oesophageal biopsies, and these proportions were not significantly different.

A total of 33 patients were diagnosed with EE, with a prevalence of 2.4%, 95% CI 1.7 – 3.4%. The prevalence was higher in the group initially referred to EGD (31, 2.3%) than in the group recruited from primary care clinics for colonoscopy and a study EGD (2, 0.1%).

Of the 33 patients with EE, 9 were diagnosed with definitive EoE, based on histology (>15 eos/hpf), presence of symptoms suggestive of oseophageal dysfunction, and the concomitant use of acid-suppressive therapy. An additional 17 patients with EE were classified as probable EoE because they met histological criteria but either were not on acid-suppressive therapy (n=4) or did not have oseophageal symptoms (n=13). Thus, the prevalence of definite EoE was 0.66% (95% CI: 0.23–1.10) and that of probable EoE was 1.25% (95% CI: 0.66–1.86). The remaining 7 patients were diagnosed with EE alone, as they had not been prescribed acid-suppressive therapy prior to endoscopy and did not have oesophageal symptoms. The two patients with EE in the primary care group had only EE (and not EoE) and neither case was diagnosed with EoE due to lack of compatible symptoms. Endoscopic findings in one patient showed two tongues of ectopic mucosa suggestive of BE (C0M1) and a 3cm hiatus hernia, and mild erosive oesophagitis in the other patient. None of the EE patients had increased eosinophils in their gastric and duodenal biopsies.

In the unadjusted analyses of possible risk factors for EE (Supplemental Table 1), a history of seasonal allergies was significantly more common in those with EE versus no EE (33% vs. 14%, p = 0.003). A lifetime of not smoking was also more common in those with EE (42.4% vs. 24.3, p = 0.02). There were no significant differences in the unadjusted analysis between the groups with and without EE with respect to age, race, history of asthma, duration of GERD symptoms, and history of peptic ulcer disease or alcohol use. Nor were there significant differences in the proportions of patients with or without EE who reported using PPI, histamine type 2 receptor antagonists or non-steroidal anti-inflammatory drugs. Neither the type nor the average dose of PPI was different between the groups (overall omeprazole once a day 20 mg in 67% and twice a day in 33%). In the endoscopy group only (Supplemental Table 2) the same determinants (seasonal allergies, non-smokers) seen for the entire group were also significant. There was only a non significant trend toward more dysphagia and less GERD in this group. The proportions of patients with GERD symptoms with or without PPI use were not different between EE and non EE (Supplemental Table 2).

Among the symptoms examined, dysphagia was significantly more common in those with EE (45.5% vs. 27.0%, p=0.03) (Supplemental Table 2). While present in some EE patients, heartburn, belching, regurgitation, nocturnal symptoms, nausea, and abdominal pain were not statistically significantly different from what they were in nonEE patients. All 11 patients with dysphagia and EE, by definition, had either definitive or probable EoE.

Among endoscopic findings, oesophageal/peptic strictures were more commonly found in those with EE (9.1% vs. 1.6%, p=0.001) than in those without EE (Table 3). Among the three patients with EE and strictures, two had definite or probable EoE. Oesophageal erosions were recorded in 15.1% of those with EE compared to 5.1% of those without EE (p = 0.01). Feline or ringed oesophagus was observed in 2 (6%) patients with EE, but this proportion was not statistically different than for EE- negative controls. Linear furrows were not observed. Endoscopically suspected BE was found in 20.5% of the total study population, including 11 of the 33 (33.3%) patients with EE. The proportion of patients with EE and endoscopically suspected BE only was significantly higher than the proportion of patients with endoscopically suspected BE without EE (15.2% vs. 6.6%, p=0.04); 7 had circumferential or maximum length < 3cm, and 4 patients had a 3-cm or longer segment (Table 3). However, there was not a statistically significant difference in definitive BE between those with or without EE. The presence of other endoscopic findings, such as, hiatal hernia, gastritis, gastric polyps, gastric or duodenal ulcer, was not statistically different between EE and non-EE patients. Although only 9.7% of patients with EE had H. pylori infection compared to 22.8% of non-EE patients, these were not statistically significant (p=0.08). Similar findings were observed in the analyses limited to the endoscopy group only, except for endoscopic and long segment BE where the association was characterized by a non significant trend (Table 3).

Table 3.

Endoscopic findings of patients with and without oesophageal eosinophilia, defined as ≥ 15 eosinophils per high-power field in oesophageal biopsies.

	Overall Study Group			Endoscopy Group Only
No. of patients	EE (+)n (%)	EE (−) n (%)	P value	EE (+)n (%)	EE (−) n (%)	P value
	33	1324		31	966
Endoscopic Findings
Duodenal Erosion	2 (6)	61 (4.6)	0.70	2 (6.5)	42 (4.4)	0.58
Oesophageal ring	2 (6)	98 (7.4)	0.80	2 (6.5)	79 (8.2)	0.76
Oesophageal erosion	5 (15.1)	68 (5.1)	0.01	5 (16.1)	48 (5.0)	0.007
Hiatal hernia	27 (81.8)	874 (65.9)	0.06	25 (80.7)	672 (69.6)	0.19
Oesophageal stricture	3 (9.1)	21 (1.6)	0.001	3 (9.68)	19 (2.0)	0.004
Gastric polyp	1 (3)	78 (5.9)	0.48	1 (3.2)	53 (5.5)	0.58
Gastric ulcer	1 (3)	24 (1.8)	0.61	1 (3.2)	17 (1.8)	0.55
Duodenal ulcer	0	8 (0.6)	0.65	0	7 (0.7)	0.63
H. pylori	3 (9.7)	285 (22.8)	0.08	3 (9.7)	177 (18.3)	0.43
BE
Definite	6 (18.2)	180 (13.6)	0.31	6 (19.4)	156 (15.5)	0.44
Endoscopic only	5 (15.2)	87 (6.6)	0.04	4 (12.9)	64 (6.6)	0.14
No BE	22 (66.7)	1057 (79.8)		21 (67.7)	752 (77.9)
BE Length*
Short segment	7 (21.2)	187 (14.1)	0.16	6 (19.4)	141 (14.6)	0.37
Long segment	4 (12.1)	78 (5.9)	0.09	4 (12.9)	71 (7.4)	0.20
No BE	22 (66.7)	1057 (79.8)		21 (67.7)	752 (77.9)

^*2 EE-negative cases were missing BE length.

EE = oesophageal eosinophilia; BE = Barrett’s oesophagus

In the multivariate analysis adjusted for age and race, the presence of seasonal allergies and current smoking remained significantly associated with EE in the analyses that included all study patients (Table 4). Oesophageal strictures was also associated with an increased risk for EE, but with only a trend toward significance (p=0.07). The diagnosis or history of H. pylori appeared to have a protective effect against EE although it did not reach statistical significance (OR 0.2; 95% CI 0.03 – 1.55, p = 0.12). Similar but non-significant trends were also observed in the endoscopy group only (Table 4).

Table 4.

Three models used to examine risk factors, symptoms and endoscopic findings as predictors for EE

	Overall Group			Endoscopy Group Only
	Odds Ratio	95% Confidence Intervals	p value	Odds Ratio	95% Confidence Intervals	p value
Risk factors*
Seasonal allergies	2.78	1.26 – 6.11	0.01	1.97	0.68 – 5.67	0.21
Asthma	1.60	0.51 – 4.97	0.42	0.79	0.10 – 6.24	0.82
Current smoker (vs. nonsmoker)	0.20	0.06 – 0.72	0.01	0.28	0.07 – 1.05	0.06
Former smoker (vs. nonsmoker)	0.68	0.31 – 1.49	0.34	0.62	0.22 – 1.78	0.38
Symptoms*
Dysphagia	1.26	0.53 – 2.99	0.60	1.28	0.50 – 3.31	0.61
Acid regurgitation in preceding year	1.60	0.63 – 4.05	0.34	1.73	0.55 – 5.43	0.35
Heartburn in preceding year	1.08	0.46 – 2.50	0.86	0.81	0.31 – 2.09	0.66
Endoscopic Findings*
Oesophageal or peptic stricture	4.50	0.90 – 22.40	0.07	4.84	0.94 – 24.9	0.06
Endoscopic Barrett’s oesophagus	1.92	0.50 – 7.28	0.34	1.23	0.25 – 6.15	0.80
Definite Barrett’s oesophagus	1.36	0.42 – 4.43	0.61	1.26	0.37 – 4.43	0.71
H. pylori	0.20	0.03–1.55	0.12	0.26	0.35 – 1.04	0.20

^*Adjusted for age and race/ethnicity

Discussion

We performed a prospective cross-sectional study of 1357 individuals, 73% consecutively presenting for elective EGD for all causes, and 27% recruited from primary care clinics for a study EGD in addition to colonoscopy. The latter group approximates the general population of patients in the VA. Thus, our report pertains to the groups least well examined to date for prevalence and risk factors of EE and EoE, namely all comers to EGD and primary care populations. All participants had at least one distal oesophageal biopsy using Jumbo forceps. A detailed reflux questionnaire was completed, and a medication and medical history were obtained prior to endoscopy. A total of 33 patients (2.4%) were diagnosed with EE, defined as ≥ 15 eosinophils per high-power field; the prevalence was 2.3% in the elective endoscopy group and a considerably lower 0.1% in the colonoscopy group. Most patients with EE (26 of 33 or 78.8%) were classified as having either definite or probable EoE, based on the presence of oesophageal symptoms and/or the use of acid-suppressive therapy at time of endoscopy; and only few patients (22.2%) had EE without EoE. Therefore, the prevalence of definite or probable EoE was 1.9% in the endoscopy group and none, 0%, in the primary care group.

EE and EoE estimates for the primary care group are likely to be indicative of the background VA patient population at large. Amongst the patients eligible for screening colonoscopy who were recruited from primary care clinic, only 2 of 360 (0.5%) were diagnosed with EE. Neither case was diagnosed with EoE due to lack of compatible symptoms. Endoscopic findings in one patient showed two tongues of ectopic mucosa suggestive of BE and a 3cm hiatus hernia, and mild erosive oesophagitis in the other patient. Thus EE could be attributed to GERD in both of these patients.

The median age of participants diagnosed with EE in the endoscopy group was 61.5 (IQR: 55.3–66.8) and for the two patients with EE in the colonoscopy group it was 65.0 (IQR: 59.2–72.7), which were reflective of the age of the underlying study groups. These ages are considerably younger than those previously reported; for example, they were 41 and 51 years, respectively in two previous similarly designed studies^{5, 17}. This highlights the fact that EoE is not confined to younger populations.

In multivariate analysis, a history of seasonal allergies, and the presence of oesophageal strictures were associated with increased risk of EE, whereas current tobacco smokers was associated with a lower EE risk. Besides a real inverse association, it is possible that the latter observation reflects confounding by GERD. Cigarette smoking promotes GERD through nicotine lowering the lower oesophageal sphincter pressure, and prolonged oesophageal acid clearance due to hyposalivation. PPI use (because of GERD) in this case will be disproportionately high and may result in reduced frequency of GERD-related EE. However, smoking appears to be protective against the development of ulcerative colitis as well as ameliorating its clinical course in some patients. It is therefore plausible that smoking through similar mechanisms blunts the inflammatory process in those predisposed to the development of EoE.

Endoscopically suspected BE was more prevalent in EE patients in unadjusted analysis but these differences did not reach statistical significance when adjusted for age and race. In multivariate analysis, the presence or history of H. pylori infection was protective against EE although this did not reach statistical significance due to the overall low prevalence of EE within the study sample. This is consistent with a previously published study revealing that EE as well as EoE was inversely associated with H. pylori infection.¹⁹

This is one of the largest prospective studies evaluating the prevalence of EE and EoE in patients presenting for non-urgent endoscopy. Most previous studies reported EoE prevalence in those with food bolus obstruction or dysphagia ^{4, 7}. Previous studies have also reported the prevalence of EoE, based mostly on histology ^6,9,10,13. The prevalence estimates of EE in our study were close to those found in a population- based study in Sweden, in which the prevalence of >15 eosinophils/hpf in at least 1 biopsy specimen was 1.1%.¹³ A review of a large pathology database of 233,649 unique patients in the United States found that 4.3% had EE (>15 eos/hpf); however, the study did not pertain to all comers to endoscopy.¹⁵

Our findings are supportive of the current cutoff of 15 per hpf, as it is not commonly found, and it is highly specific for EoE; since most patients with EE, according to that definition, had EoE. The findings are also not supportive of routine oesophageal biopsies for the purpose of detecting EE. This rationale is further supported by a retrospective review of histopathology and endoscopy records in Calgary, Alberta, Canada which revealed 92% of patients with histological evidence of >15 eos/HPF had dysphagia or food bolus impaction.²⁴ Most patients diagnosed with EE can be classified as having EoE, given the high rate of oesophageal symptoms among these patients. These patients would have been diagnosed by targeted screening in patients with symptoms.

Although definitive BE was also more common in EE patients than in those without EE, these difference were not significant, likely because of the small numbers. There have been recent reports of new-onset EE among patients with BE post radiofrequency ablation.^20,21 While it is possible that healing of EE or EoE can result in the development of BE, given the cross-sectional nature of our study, we are unable to discern the temporal association of these two conditions

Our study evaluated all comers to EGD (not just those with dysphagia) and defined EoE as >15 eos/HPF with oesophageal dysmotility symptoms and concomitant PPI use at the time of endoscopy and found that most patients with EE have EoE. Differences in the sampling frame and study design are likely to explain the differences in findings between our study and previous studies. Prasad et al.⁵ prospectively assessed using midoesophageal biopsies the prevalence of EE in patients with dysphagia and no endoscopic abnormality. EE -defined as the presence of >20 eosinophils/hpf but not non-response to PPI- was found in only 8 of 21 (38%) patients with endoscopic changes suggestive of EE with the rest presumably due to GERD. However, the lack of clear distinction between EE and EoE, the selection bias, and the absence of non response to PPI in the EE definition are the major differences from our study and these issues would explain the relatively high GERD among patients with EE in their study. Molina Infante et al.¹⁷ reported a prospective study of consecutive adult patients referred for EGD due to upper gastrointestinal symptoms off PPI for at least 1 week prior, in whom a 4 oesophageal biopsies were taken 10 cm above the oesophagogastric junction. EoE was defined by persistence >15 eos/HPF after treatment with twice daily PPI for 8 weeks. The study reported EoE in 1.26% (9/712) of their study subjects. However, only approximately 25.7% (9/35) with EE were diagnosed with EoE. The difference from our study that EGD was performed off PPI while in our study approximately 70% were on PPI. Lastly, Rodrigo et al.²³ reported the findings of a retrospective pathology database study of 3,648 patients who underwent EGD with biopsy for any indication between 2002 and 2005. A total of 40 patients were found to have >20 eos/hpf, 32 of whom underwent 24 hour pH testing while off medications and 20 of whom had abnormal GERD. The lack of information on PPI use at the time of endoscopy is the major difference from our study. If most patients with >20 eos/hpf had endoscopy while on PPI therapy (as would happen in practice) and subsequently underwent pH testing off acid suppressive medication this will overestimate EE cases attributed to GERD. PPI use in our study population may have resulted in less EE among patients with GERD thus leaving the rest of the EE to qualify as probable or definite EoE. This however closely mirrors what is done in most clinical practices.

It is possible that inadequate oesophageal biopsies led to an underestimation of prevalence of EoE in this study. Only one oesophageal biopsy was taken in most of our study subjects. It is possible that standardized additional biopsies could have increased the detection of EE. Practice guidelines are not specific about the number or locations of biopsies to obtain ^{3, 18}, because there is no evidence from prospective studies to support or quantify the additional EE yield associated with more biopsies. Nevertheless, a retrospective study by Gonsalves et al. ²² found that while 5 biopsies had a sensitivity of 100% for finding >15 eos per HPF, 1 biopsy had a sensitivity of only 55%. Syed et al.²⁴ also found that the incidence of >15 eos/HPF increased in a Canadian population during a 5 year time period during which the rates of esophageal biopsy increased per endoscopies performed. This finding was even more pronounced in those undergoing endoscopy for dysphagia or food bolus impaction. Several factors may mitigate the effect of this limitation: our biopsies were obtained using Jumbo forceps and, therefore, were more representative than those obtained with a regular forceps; 11.7% (4/34) of EE cases and 7.0% of controls had two samples or more, and additional oesophageal biopsies were also taken from 20.5% with suspected BE. Furthermore, the proportion of EE in our study (2.4%) is within the range reported in two other prospective studies of all comers for EGD (1.1%¹³ and 6.5%¹⁰) that used only one distal oesophageal biopsy¹³ and 4 proximal and 4 distal oesophageal biopsies¹⁰, respectively. Lastly, a large retrospective pathology database review found the prevalence of >20 eos/HPF to be 1.1% with one biopsy taken the distal oesophagus.²³

We did not perform reflux monitoring to document GERD; so it is possible that some patients with non EoE-related EE had GERD. This potential misclassification would reduce the prevalence of definite or probable EoE in the study. Furthermore, the use of PPIs in this study population was common; and PPIs have been shown to be effective in achieving remission of oesophageal eosinophilic infiltration in some patients¹⁷, so it is possible that the prevalence of EE is higher in a non-PPI-treated population. Nevertheless, in our stratified analysis, there were no significant differences in EE prevalence between PPI users and non-users.

In summary, this study shows that EE defined by ≥ 15 eosinophils per high-power field is infrequent among patients without EoE. Routine oesophageal biopsies cannot be recommended in patients without symptoms related to EE.