WHAT IS ALREADY KNOWN ON THIS TOPIC

Eosinophilic oesophagitis (EoE) is a leading cause of food impaction and dysphagia in children and adults.

Current guidelines recommend treatment with proton pump inhibitors, swallowed topical corticosteroids, and oral corticosteroids, in addition to food elimination diets.

WHAT THIS STUDY ADDS

Time to diagnosis of EoE was prolonged in patients with acid reflux/heartburn, vomiting and nausea, non-cardiac chest pain, ≥1 EoE-related comorbidity or age <18 years.

Patients who developed strictures were more frequently men, and reported dysphagia and food impaction at baseline.

HOW THIS STUDY MIGHT AFFECT RESEARCH PRACTICE OR POLICY

This study provides an insight into the real-world experiences and challenges of diagnosing and managing patients with EoE, and could help practitioners diagnose and avoid complications in their patients.

Introduction

Eosinophilic oesophagitis (EoE) is a chronic, progressive, immune-mediated disease, characterised by symptoms of oesophageal dysfunction and eosinophil infiltration into the oesophageal tissues.1 If left untreated, EoE can progress to oesophageal remodelling, rigidity and luminal narrowing due to strictures.2 3

The prevalence of EoE varies by country and is estimated to be 13–49/100 000 people globally4 5 and 0.2–43/100 000 children (≤18 years).6 Common symptoms include dysphagia, vomiting, feeding difficulties (in children), food impaction and chest and abdominal pain.4 7–12 Frequently reported comorbidities include allergic rhinitis, asthma, pollen food allergy syndrome, drug allergy, food allergy and psychiatric disease.13 14

Treatment typically consists of swallowed topical corticosteroids (TCS), proton pump inhibitors (PPIs), dietary therapies and procedures such as endoscopic dilation.4 15 TCS have demonstrated greater effectiveness than PPIs at inducing clinical remission16 and are recommended over the use of oral corticosteroids (OCS), which are used sparingly to minimise associated side effects.17 18

Early diagnosis and treatment are key to minimising disease progression and associated complications.19 20 Understanding an individual’s risk of disease progression could better inform treatment discussions. The purpose of this study was to provide up-to-date data concerning demographics and treatment patterns in patients with EoE in England. Potential relationships between patient characteristics and prolonged time to diagnosis or the development of strictures were also explored.

MethodsObjectives

The primary study objectives were to describe the demographic characteristics, clinical symptoms and comorbidities (asthma, eczema, allergic rhinitis and food allergy/oral allergy syndrome) in patients with EoE and in a non-EoE general population cohort and to describe treatment patterns in patients with EoE.

The exploratory study objectives included: describing the time to diagnosis in patients with EoE and exploring the relationship with patient characteristics; describing the characteristics of patients who develop strictures and describing the proportion of patients achieving adequate disease control.

Study design

This was an observational, retrospective cohort study of patients with incident EoE. Routinely collected data were sourced from three linked databases in England: the Clinical Practice Research Datalink (CPRD) Aurum database; the Hospital Episode Statistics (HES) database; and the Office for National Statistics (ONS) mortality files. The CPRD is a large database containing anonymous longitudinal primary care medical records of patients registered at primary care practices in England. It contains information on primary healthcare utilisation, including outpatient prescription medications at patient level and has been shown to be of a high quality in a number of validation studies.21 The HES provides information on hospital admissions, accident and emergency attendance and outpatient visits across National Health Service hospitals in England and secondary care data including treatment characteristics, disease severity and service use. The ONS mortality files report death registration data to identify mortality and cause of death. Linkage, by patient identifiers held by CPRD, was conducted by CPRD.

EoE study population

Paediatric (<18 years) and adult (≥18 years) patients in the CPRD, who could be linked to HES data, had ≥1 diagnosis of EoE recorded using a diagnostic code (online supplemental table 1) during the study period (January 2014–October 2020) and had ≥12 months (baseline period) of continuous registration with the practice before their first EoE diagnosis, were eligible for inclusion.

Patients were followed from date of first EoE diagnosis (index date) until the earliest of the following: patient death, deregistration from their general practitioner (GP) practice or the last CPRD data collection date. A follow-up period of ≥12 months was required for inclusion in the analyses of treatment type, development of strictures and time from adequate to poor disease control. A ≥3-year follow-up period was required for the assessment of changes in treatment over time to ensure patterns were captured adequately. Additionally, ≥3 years of registration with the GP practice prior to first diagnosis was required for inclusion in the time to diagnosis analyses.

A matched non-EoE, general population reference cohort was used to generate reference data for clinical characteristics. Age-matched and sex-matched controls were randomly selected from the CPRD database and comprised individuals who had no documented diagnosis of EoE in the database and who had ≥12 months of registration with the same GP practice before the matched index date.

Reference patients were assigned the index date of the matched EoE patient without replacement. Exact matching on a 1:4 ratio of patients with EoE to matched controls without EoE was performed for sex and GP practice. Patients were also matched for age (year of birth), duration of look-back period prior to index date and duration of follow-up after index date.

Statistical analysis

Demographic characteristics, clinical symptoms and comorbidities for the EoE cohort and matched controls were reported descriptively as percentages of categorical variables and as mean (SD) and median (IQR) of continuous variables, depending on the distribution of data. P values for differences between groups were calculated using Pearson’s χ2 test for proportions, two-sided t-test for mean values and Mann–Whitney U test for median values, with a cut-off of p<0.05. The use of TCS, PPI, OCS, biologics, immunomodulators, antihistamines, antileukotrienes, mast cell stabilisers, histamine-receptor antagonists, dietary therapies and monitoring or procedural intervention at any timepoint were reported descriptively. Switches between treatments and treatment continuation/discontinuation were reported. Time to diagnosis, defined as the time between first recorded EoE symptom in the database (online supplemental methods) and EoE diagnosis, was summarised as mean (SD) and median (IQR). Patient demographics, clinical symptoms, disease severity, comorbidities and procedures were described by time to diagnosis interval (no delay, >0–<12, 12–<24 and 24–36 months). In addition, unadjusted and adjusted incident rate ratios were calculated using negative binomial regression models to evaluate disease severity, upper gastrointestinal (GI) endoscopy and oesophageal dilation as potential determinants of time to diagnosis. Rate ratios were adjusted using the matched cohort set for age, sex, body mass index, ethnicity and the following comorbidities measured during baseline: concomitant allergic diseases (including asthma), oesophagitis and depression. Statistical significance was determined using 95% CIs and respective Wald test (p<0.05). Patient characteristics were described for those who developed strictures and those who did not.

Patients were classed as having severe disease if they had undergone endoscopic dilation, were prescribed a systemic corticosteroid or had >1 hospital visit for food impaction or endoscopic bolus removal over 12 months.

Adequate and poor disease control are defined in the online supplemental methods. The number of patients achieving ≥3 months of adequate disease control were reported, and time from adequate to poor disease control was summarised as median (IQR). Patient characteristics were described for those who moved from adequate to poor disease control at any timepoint.

ResultsBaseline demographics

The study included 2381 patients with incident EoE (online supplemental figure 1). Overall, 70.1% were men; median (IQR) age was 40 (25) years and 14.2% of patients were aged <18 years. Similar results were observed for the 9365 matched controls (table 1, online supplemental table 2). The majority of those who reported ethnicity were white, for both the EoE cases (88.6%) and matched controls (85.7%).

Table 1

Baseline characteristics for patients with EoE and non-EoE matched controls in England

Reports of ≥1 EoE-related symptom during the baseline period were significantly higher (p<0.001) in patients with EoE (66.4%) compared with matched controls (6.7%). The most frequently reported symptoms in patients with EoE were dysphagia, acid reflux/heartburn (including gastro-oesophageal reflux disease), abdominal pain and food impaction (table 1). The frequency of dysphagia, food impaction and non-cardiac chest pain increased with increasing age, whereas regurgitation, vomiting and nausea were more frequent in children than adults (online supplemental table 2).

Patients with EoE reported ≥1 allergic comorbidity more frequently than matched controls (29.2% vs 9.1%; p<0.001), with asthma the most commonly reported (23.8% of patients with EoE; online supplemental table 3). The frequencies of atopic dermatitis and food allergy decreased with increasing age (online supplemental table 2). The most frequently reported EoE-related comorbidity was reflux oesophagitis (20.5% of patients with EoE). Depression and/or anxiety were reported for 9.4% of patients with EoE and 5.2% of matched controls. Severe EoE was reported for 15.4% of patients.

Treatment patterns

A total of 1919 (80.6%) patients with EoE had ≥12 months of follow-up (mean (SD) 1132 (530) days) (online supplemental figure 1). Most (90.1%) of these patients were prescribed a PPI, TCS or OCS at any timepoint, and a small proportion was prescribed food formulas or elimination diets (exclusion of specific food groups from the diet) (table 2). Most (82.9%) reported an upper GI endoscopy during the baseline or follow-up periods; 6.6% underwent endoscopic bolus removal and 4.1% underwent oesophageal dilation at any timepoint.

Table 2

Proportion of patients in England prescribed PPIs, TCS or OCS and other treatments at baseline, first 12 months and ever (baseline or during follow-up)

There were 917 patients with EoE who had ≥3 years of follow-up data available for the assessment of treatment patterns over time (mean (SD) 1595 (360) days). The most common first-line therapy was PPI monotherapy, and the most common therapy patients switched to second line was PPI plus TCS (figure 1). A small proportion of these patients (7.6%) were not prescribed drug therapy during follow-up. Nearly one-third (29.2%) remained on first-line therapy; 21.8% discontinued and did not switch to another drug, and 41.3% of patients switched to second-line therapy. Of those who switched, 23.0% remained on second-line therapy, 6.8% discontinued second-line therapy and did not switch to another drug, and 11.6% switched to third-line drug therapy, mainly PPIs.

Figure 1

Overview of the treatments received in all patients with EoE in England with at least 3 years of follow-up (n=917). EoE, eosinophilic oesophagitis; OCS, oral corticosteroids; PPI, proton pump inhibitor; TCS, topical swallowed corticosteroids.

Time to diagnosis

Overall, 2015 patients were registered with their GP for ≥3 years before first EoE diagnosis. Of these, 1473 patients experienced symptoms of EoE (dysphagia, first food impaction, acid reflux/heartburn, non-cardiac chest pain or failure to thrive) reported on the same day as or before EoE diagnosis. Median (IQR) time to diagnosis was 4.5 (1.3 to 16.4) months (mean (SD), 9.8 (10.9) months). Time to diagnosis was <12 months for the majority (68.2%), 12–<24 months for 16.3% and 24–36 months for 15.5% of these patients (online supplemental table 4).

Patients with a prolonged time to diagnosis (24–36 months) reported increased acid reflux/heartburn, vomiting and nausea, non-cardiac chest pain, ≥1 EoE-related comorbidity and age <18 years compared with those with a time to diagnosis of 0, >0–<12 and 12–<24 months. Food impaction was reported more frequently by patients with no delay to diagnosis than patients with prolonged time to diagnosis (figure 2; online supplemental table 4).

Figure 2

Characteristics observed with increasing duration of time to diagnosis (0, >0–<12, 12–<24, 24–36 months) in English patients with EoE with data available at least 3 years prior and including the EoE diagnosis date (n=1473). EoE, eosinophilic oesophagitis.

Patients with severe disease had a shorter time to diagnosis than those with non-severe disease, after adjusting for covariates (adjusted incident rate ratio 0.77; 95% CI 0.65 to 0.91). A non-significant increase in time to diagnosis was seen in patients requiring upper GI endoscopy (compared with no upper GI endoscopy) during the baseline period (online supplemental table 5). There was no association between oesophageal dilation during baseline and time to diagnosis. Similar results were obtained in sensitivity analyses that excluded patients diagnosed with first EoE symptoms on the same day as EoE diagnosis.

Development of strictures

Among 1919 patients with EoE with ≥12 months of follow-up, 18.3% developed strictures at any timepoint. During the baseline period, acid reflux/heartburn, vomiting, nausea, abdominal pain and age <18 years were reported less frequently and severe disease, male gender, dysphagia and food impaction, more frequently, in those who developed strictures than those who did not (online supplemental table 6).

Disease control

All patients who initiated PPIs/TCS/OCS during the baseline or follow-up periods (N=1729) achieved adequate disease control for ≥3 months following treatment initiation; 61.7% of these patients moved to poor disease control during follow-up. The median (IQR) time from adequate to poor disease control was 146 (305.5) days. Compared with patients who remained in adequate disease control, those who moved from adequate to poor disease control at any point in the study more frequently reported severe disease, ≥1 clinical symptom, ≥1 allergic comorbidity, ≥1 EoE-related comorbidity, ≥1 psychiatric comorbidity or prescription of ≥1 psychiatric medication during baseline (online supplemental figure 2).

Discussion

This retrospective observational study described EoE symptoms, comorbidities and treatment patterns in children and adults with EoE in England. EoE-related symptoms and comorbidities during baseline were reported more frequently in patients with EoE than in matched controls, the most common symptoms being dysphagia, acid reflux/heartburn, abdominal pain and food impaction. The most common EoE-related comorbidity during the baseline period was reflux oesophagitis, although this may be attributable to incorrect reflux diagnoses in patients with EoE owing to heartburn symptoms associated with the disease.22 23

Most patients received a PPI, TCS or OCS. The most common first-line therapy was a PPI, reflective of current guideline recommendations and similar to treatment patterns previously reported in the USA.4 24 One-third of patients remained on first-line therapy without switching. The proportions of patients undergoing endoscopic bolus removal or oesophageal dilation at any timepoint were low (6.6% and 4.1%, respectively) compared with those reported in other studies (16%–35% and 23%, respectively); this may have been due to differences in study design such as longer follow-up periods (up to 18 years) and the involvement of different healthcare systems.24–27

The time to diagnosis reported for patients included in our study (mean (SD), 9.8 (10.9) and median (IQR) 4.5 (1.3 to 16.4) months) was shorter than that reported in other studies: a prospective database study of Swiss patients reported a median time to diagnosis of 4 years, with 32% having a time to diagnosis ≥10 years28; while a retrospective study in North Denmark reported a mean (SD) time to diagnosis of 10 (12) years.29 These differences may be due to variations in study design, for instance, our study required ≥3 years of baseline data for inclusion in the time to diagnosis analysis. It is also important to note that the definition of time to diagnosis in this study was reliant on patients reporting symptoms to their primary care provider. It is possible that patients may have experienced symptoms sooner but were unaware that these were a manifestation of EoE. Therefore, time to diagnosis in this instance may be an under-representation of when patients truly started to experience symptoms. Thus, it may also be that patients with EoE who did not feel their symptoms were severe enough to see a healthcare provider were not captured in this study.

Overall, one-third of patients in this study had a time to diagnosis of ≥12 months, highlighting the need for earlier diagnosis to minimise disease progression. In our study, patients with a prolonged time to diagnosis (24–36 months) were more frequently of a younger age (<18 years) than those reporting shorter times to diagnosis. Non-specific GI symptoms, such as abdominal pain, acid reflux and vomiting, are more common in children and may contribute to the delayed diagnosis for this age group.30 Additionally, patients with no diagnostic delay reported severe disease more frequently than patients with prolonged time to diagnosis.

Nearly one-fifth of all patients reported stricture development during the follow-up period. Severe disease, male gender and the presence of dysphagia and food impaction during the baseline period were more frequently reported among patients who developed strictures than those who did not. While food impaction and dysphagia are often reported as risk factors for stricture,31 32 there are conflicting findings among studies investigating gender-based differences in the presentation of EoE.33

Data from the CPRD Aurum database cover 20% of the English population and are broadly representative of England’s wider population regarding age, gender and ethnicity.34 35 Accordingly, a study strength is that the findings are generalisable to patients with EoE treated in primary care in England. Linking CPRD with HES provides comprehensive data on hospital care and a more complete view of the patient journey. Moreover, the long follow-up period improves real-world understanding of longitudinal treatment patterns. The limitations of this study were common to retrospective observational studies using routinely collected electronic health record data. Electronic health records do not directly report patient symptoms and are dependent on healthcare professionals entering data into the record. These data may be missing or include misclassification or inconsistencies in coding within and between practices and over time. EoE case ascertainment required at least one diagnostic code recorded in the CPRD Aurum database, which relied on physician reporting of the diagnosis. Furthermore, data from the CPRD Aurum database are deidentified, so it was not possible to confirm symptoms with patients retrospectively. Information on secondary care prescriptions (eg, biologics) and over-the-counter use (in particular, PPIs such as omeprazole) are not recorded in primary care and may have been missed in our dataset. Dietetic outcomes other than elimination diets and formula prescriptions were not captured. The low proportion of patients on elimination diets and food formulas suggests such diets are not well recorded in the CPRD, hence, this study did not provide a complete picture of dietary therapies. Reasons for this may include under-reporting of dietary therapies by physicians, over-the-counter use not recorded and patients following exclusion diets independently, without prescription. Finally, as there are currently no internationally accepted definitions of disease control, a proxy was developed that defined disease control in terms of changes in resource use. Disease control could not be measured using changes in symptoms and/or histological response due to the retrospective nature of the study, which did not enable histological evaluation as a marker of disease control.

Conclusion

This study provides valuable insights into the real-world experiences and challenges of managing patients with EoE. These findings may help healthcare professionals in identifying patients with EoE who require closer monitoring to avoid complications and highlight the need for earlier diagnosis and treatment.

Ethics statementsPatient consent for publication

Not applicable.

Ethics approval

This research was conducted in full conformance with ethical principles that are consistent with the Declaration of Helsinki, ICH GCPs, and the Guidelines for Good Pharmacoepidemiology Practice (GPP) published by the International Society of Pharmacoepidemiology (ISPE), and appropriate laws and regulations of the UK. All data within the dataset provided by CPRD Aurum and HES are fully anonymised. This study protocol was submitted to the Independent Scientific Advisory Committee (ISAC) for approval.

Acknowledgments

This manuscript was supported by AstraZeneca. This work uses data provided by patients and collected by the NHS as part of their care and support. The authors would like to thank Steady Chasimpha and Medha Strivastava for statistical analysis. The authors would like to thank Juliette Gray, PhD, and Hannah Talbot, BSc, of inScience Communications, Springer Healthcare Ltd, UK, for providing medical writing support, which was funded by AstraZeneca in accordance with Good Publication Practice (GPP3) guidelines (http://www.ismpp.org/gpp3).