Systematic Review With Meta-analysis

The Growing Incidence and Prevalence of Eosinophilic Oesophagitis in Children and Adults in Population-based Studies

Pilar Navarro; Ángel Arias; Laura Arias-González; Emilio J. Laserna-Mendieta; Miriam Ruiz-Ponce; Alfredo J. Lucendo


Aliment Pharmacol Ther. 2019;49(9):1116-1125. 

In This Article


This systematic review provides a detailed update of the existing data on the worldwide incidence and prevalence of EoE. Comparison between studies was attempted in relation to geography and age. In addition, we investigated whether observed differences in disease occurrence and changes over time might have been due to varying definitions for the disease. The results from the systematic review confirmed that EoE is a common disease, less common in children than in adults; and identifies its growing frequency.

Prevalence rates for EoE showed a worldwide geographical variation, which probably reflects the fact that EoE is directly related to socio-demographic development and the recognition of EoE in developed countries. The higher reported prevalence of EoE was in North America and Europe, with most of the studies coming from these countries. Differences in prevalence among studies are widely related to varying case definition (with evolving diagnostic criteria for EoE in the last decade) rather than solely geographic variation.

The main purpose of this investigation, and also the main difference from other reviews, was to analyse data on the general population only and to cluster studies according to their focus, whether adults or children. We also aimed to update the figures provided by our previous systematic review, which summarised publications up to the end of 2014;[5] with the expanding recognition of EoE, a sharp increase in numbers of reported cases, together with the 17 population-based studies released since 2014/5 (which represent half of the documents included in this systematic review), determined that previous results were obsolete. Furthermore, the recent update of diagnostic criteria for EoE, by eliminating proton pump inhibitor-responsive oesophageal eosinophilia (PPI-REE) as a different entity and including these patients within the clinical spectrum of EoE,[1,6,45] required an update of the data in order to reflect the magnitude of the problem, especially since several recent population studies had already applied the new diagnostic criteria.[9,40,7,44]

We are able to confirm the findings from our previous systematic review showing that EoE overall is less common in children than in adults.[5] It is worth noting that there has been a significant increase in the overall prevalence of EoE for both age groups since the summary estimates provided up to the end of 2014 were documented: prevalence rates in children increased from 19.1 (95% CI, 7.9-35.2) to 34.4 (95% CI, 22.3-49.2) patients per 100 000 inhabitants in less than 4 years, while for adults they grew from 32.5 (95% CI, 12.4-62.2) to 42.2 (95% CI, 31.1-55) patients per 100 000 inhabitants. The rising prevalence of the disease cannot be attributed only to the accumulation of cases over time, but also to a continuous and ongoing increase in incidence rates. Thus, the overall incidence rate of EoE increased from 3.7 (95% CI, 1.7-6.5) to 4.4 new cases/100 000 persons-year (95% CI, 2.8-6.4) in less than 3 years, with increases both for children (from 5.1; 95% CI, 1.5-10.9 to 6.6; 95% CI, 3-11.7) and for adults (7; 95% CI, 1-18.3 to 7.7; 95% CI, 1.8-17.8), with even higher rates provided by the most recent studies. The reasons behind this increase have not been clarified completely and are urgently needed. It has been argued for example, that most of the previous population-based studies underestimated the magnitude of EoE by excluding patients with a response to PPIs.[41] Only the most recently published papers included in our review had EoE diagnosed by the current evidence-based criteria,[6,7] according to which, a response to PPI does not preclude a diagnosis of EoE, contrary with previous consensus guidelines.[2,18] However, multiple studies, both in the early literature[19–21,43] and in that published after the proposal of the so called PPI-REE in 2011,[31,36,42] did not exclude response to PPIs as a diagnostic requirement for EoE. In any case, it is clear that the frequency of EoE has progressively increased over the years as the criteria for the disease have been updated, reflecting a change in the epidemiology of the disease beyond that of including patients who previously responded to PPIs within the epidemiological calculations.

A more widespread, general use of endoscopy for the diagnosis and management of gastroenterological disorders was also proposed as an explanation for the increasing frequency of EoE, together with a greater awareness by clinicians that now consider EoE within the differential diagnosis of oesophageal dysfunction symptoms.[46,47] However, recent studies have demonstrated that the increase in new EoE cases goes beyond the use of endoscopy with biopsy,[41,7,48] thus supporting true expansion in the epidemiology of the disease in several settings.

Our review has several strengths, such as: compiling results from an exhaustive literature search of three major databases with no time limit on publication date; critically appraising the studies recovered according to their methodology and risk of bias; different investigators independently extracting the data from the studies which were included; and risk of bias being assessed with a validated document developed for prevalence studies[13] by two independent researchers. We also assessed the potential effect of the changing definition of EoE over the last decade on the epidemiology of the disease, and the accuracy of data obtained from administrative and insurance databases in relation to those extracted from hospital-based case series.

However, some limitations should also be acknowledged, including the possibility of not recovering all the relevant information published on population-based epidemiological data concerning EoE, despite our attempts to minimise this risk. This fact could also affect the lack of significant publication bias found in our Funnel plot analyses. Furthermore, most of the information retrieved comes from retrospective registries of codified diagnoses, and the reliability of this information was not systematically checked. Finally, a high I 2 value may identify estimates with low predictive values, thus limiting the reliability of the results of our meta-analyses.

In conclusion, our results confirm that EoE currently constitutes a highly prevalent disorder, with rising incidence and prevalence rates in recent years. The increasing frequency of EoE overall in population-based studies, which are mainly restricted to US and Europe, have consistently demonstrated the predominance of EoE among adults compared to children. The high, and still ongoing prevalence of EoE in developed countries, should prompt resources to be allocated in order to face the costs associated with the diagnosis and treatment of EoE, and to design sustainable health policies with regard to the chronic nature and impact of the disease on patients' health.