Risk factors and 26-years worldwide prevalence of endoscopic erosive esophagitis from 1997 to 2022: a meta-analysis

Erosive esophagitis (EE) is the part of gastroesophageal reflux disease (GERD) spectrum and may progress to esophageal adenocarcinoma. Due to its progressivity and unclear prevalence, we aim to identify the factors contributing in EE to decide the need for further examination. We performed a PRISMA 2020-based systematic search through PubMed and other resources up to June 2, 2022. Study quality was assessed using the Newcastle–Ottawa Scale (NOS). The odds ratio (OR) of each factor and worldwide prevalence of EE were measured. There are 114 observational studies included with a total of 759,100 participants. Out of 29 factors, the significant risk factors are age ≥ 60 y.o. (OR 2.03 [1.81–2.28]), White/Caucasian (OR 1.67 [1.40–1.99]), unmarried (OR 1.08 [1.03–1.14]), having GERD ≥ 5 years (OR 1.27 [1.14–1.42]), general obesity (OR 1.78 [1.61–1.98]), central obesity (OR 1.29 [1.18–1.42]), diabetes mellitus (DM) (OR 1.24 [1.17–1.32]), hypertension (OR 1.16 [1.09–1.23]), dyslipidemia (OR 1.15 [1.06–1.24]), hypertriglyceridemia (OR 1.42 [1.29–1.57]), hiatal hernia (HH) (OR 4.07 [3.21–5.17]), and non-alcoholic fatty liver disease (NAFLD) (OR 1.26 [1.18–1.34]). However, H. pylori infection (OR 0.56 [0.48–0.66]) and atrophic gastritis (OR 0.51 [0.31–0.86]) are protective towards EE. This study demonstrates that age, ethnicity, unmarried, long-term GERD, metabolic diseases, HH, and NAFLD act as risk factors for EE, whereas H. pylori infection and atrophic gastritis act as protective factors. These findings may enable a better understanding of EE and increase greater awareness to address its growing burden.

Chung H et al. 75 Seoul, Korea Retrospective cohort study LA classification 53.5 ± 10.9 Medical check-up 639 Chung TH et al. 76 Ulsan, South Korea Cross-sectional study LA classification 50.91 ± 6.4 Shipyard male workers undergoing medical check-up 5510 Chung TH et al. 14 Seoul, South Korea Retrospective cohort study LA classification 50.13 ± 6.51 Medical check-up 6874 Deppe et al. 77 Munich, Germany Cross-sectional study N/A 49.7 ± 15.1 Diagnosed with GERD   42   El-Serag et al. 78 All VA hospital in USA Case-control study N/A 58.00 ± 14.18 N/A 92,860 101,366 El-Serag et al. 79 Houston, USA Cross-sectional study LA classification 44.61 ± 10. 22  Hospital employee undergoing medical check-up 124 El-Serag et al. 80 Houston, Texas, US Cross-sectional study LA classification 44 ± 10 Hospital employee undergoing medical check-up 152 Filiberti et al. 81  www.nature.com/scientificreports/removing one study, either Chung et al. 14 , Kulig et al. 13 , Mun et al. 15 , or Sadiku et al. 16 , for marital status, and Kulig et al. 13 for disease duration.The sensitivity analyses of the remaining demographical factors suggest that the pooled effects are not influenced by any single study.After removing a study by Cho et al. 17 in the sensitivity analysis of duodenal ulcer, the pooled effect is shifted from nonsignificant to significant.On the contrary, the pooled OR of atrophic gastritis is shifted from significant to nonsignificant following the removal of a study by Ko et al. 18 .The leave-one-out sensitivity analyses of the remaining comorbidities suggest that the provided overall effects are robust and not affected by any single study.

Medication history.
We include five pharmacological medications: Non-steroidal anti-inflammatory drug (NSAID) only, aspirin only, NSAID and/or aspirin, proton pump inhibitor (PPI), H2 receptor antagonist (H2RA), and antacids (Table 2).The forest and funnel plots are provided in Supplementary Fig. S24-S29 online.NSAID and/or aspirin (I 2 = 58%) and H2RA (I 2 = 53%) have moderate heterogeneity, while PPI (I 2 = 93%) and antacids (I 2 = 84%) have high heterogeneity.All heterogeneity tests are performed using REM.There is no medication history considered as risk nor protective factors in the current analysis: NSAID only (OR 1.The sensitivity analysis of the antacids use reveals that the overall effect is changed from nonsignificant to significant following the removal of one study by Kang et al. 19 .On the other hand, no study has a notable influence in the leave-one-out sensitivity analyses of the remaining medication histories, proving the robustness of the pooled results.

Publication bias and quality assessment.
The funnel plots of central obesity (Supplementary Fig. S10B online), high LDL-C (Supplementary Fig. S16B online), and low HDL-C (Supplementary Fig. S17B online) show an asymmetrical distribution of studies, revealing the potential of publication bias.These findings are further confirmed by significant Egger's test result in each factor (Z = 2.03 and p = 0.04 for central obesity, Z = 2.16 and www.nature.com/scientificreports/p = 0.03 for high LDL-C, Z = -2.23 and p = 0.03 for low HDL-C).On the contrary, no potential of publication bias is found in the rest of the factors since their funnel plots show a rather symmetrical distribution of studies, further supported by their insignificant Egger's test results (Table 2).The quality of each study is shown in Table S1-S3.The overall quality of the included case-control studies (Supplementary Table S1 online) is good in 27 studies, while the rest (n = 9) is moderate.Of the 17 cohort studies, thirteen and four studies have good-and moderate-quality, respectively (Supplementary Table S2 online).The qualities of 61 cross-sectional studies (Supplementary Table S3 online) are as follows: (1) very good for 37 studies; (2) good for 19 studies; and (3) satisfactory for 5 studies.There are no poor-quality and unsatisfactory studies in the current meta-analysis.

Discussion
To the best of our understanding, this meta-analysis is the first to thoroughly analyze the risk factors and prevalence of EE across the world from 1997 to 2021.Our results indicate that several demographical factors-age ≥ 60 y.o., White/Caucasian, single or unmarried, and having GERD ≥ 5 years-increase the risk of having EE.Interestingly, we find both risk and protective factors towards EE in the comorbidities.Obesity, DM, hypertension, dyslipidemia, hypertriglyceridemia, HH, and NAFLD are found to increase the risk of EE, while H. pylori infection and atrophic gastritis are found to be protective towards EE.Our results also indicate that medication history Table 3. Worldwide pooled prevalence of EE based on geographical regions and countries.CI, confidence interval; EE, erosive esophagitis; HCI, higher confidence interval; LCI, lower confidence interval.www.nature.com/scientificreports/ is not significantly increasing the risk nor protective of EE.The prevalence of EE in each of America, Africa, and Europe is higher than that in Asia and the highest prevalence is found to be in Africa and the Middle East.
Our study indicates that the risk of EE in males is twice than that in females.Previous studies have suggested that the combination of behavioral, immunologic, and metabolic aspects, especially in men, can increase the risk of EE and affect its prevalence.For example, Erol and Karpyak 20 and Matsuzaki et al. 21suggest that cigarette smoking and alcohol consumption are more common in men and may increase the risk of having EE in men, approximately two to three times more than women.A longitudinal study by Adachi et al. 22 also indicates that the prevalence of EE in men during 10-year period is increasing mainly due to aging, high BMI, and large diaphragmatic hiatus.This change, however, is not found in women.Furthermore, previous studies by Yoon Kim et al. 23 and Sun Kim et al. 24 suggest the protective effects of estrogen, although the studies use animal models.
Our study shows that the risk of EE in the Western (White/Caucasian) population is approximately two-fold higher than that in the non-White/Caucasian population.Previous studies have suggested that lifestyle factors, anatomical, and genetic variance can also explain the high risk of EE in the Western population.In terms of lifestyle factors, Wirth et al. 25 , Abraham et al. 26 , and Ko et al. 18 indicate the differences in the risk can be attributed to the differences in eating habits or cultures (e.g.high fat diet and alcohol drinking in the Western population), distribution of visceral fat tissues, and body composition between the Western and Eastern populations 25,26 .In terms of anatomical differences, previous studies also suggest that the mass of gastric parietal cells of Western population is greater than that in the Asian population, which explains the higher gastric acid production in the Western population 18,25 .Moreover, in terms of genetic variance, some previous studies indicate that the difference in the ABH-secretor and Lewis histo-blood group may explain the difference of risk in the Western population.In particular, Wirth et al. 25 and Suzuki et al. 27 indicate that individuals with group A and non-secretors (common in the Western population) are prone to have EE.
This study finds that HH increases the risk of EE and this may be explained by anatomical and physiological factors.HH may diminish the augmenting effect of diaphragmatic crus to prevent gastric reflux 28 .Previous study mentions that the size of the HH is the most important risk factor of EE in individuals with GERD 29 .Some etiologies, such as pregnancies, surgical history, being elderly, and overweight, may increase the probability of HH 30,31 .
Obese individuals tend to experience more frequent and intense reflux symptoms compared to non-obese individuals 32 .Anatomically, obesity may promote esophagitis by increasing intra-abdominal pressure (IAP) and inducing lower esophageal sphincter (LES) relaxation 33 .Another evidence also reveals that obesity increases the transvesically-measured IAP 34 .Another mechanism thought to be involved in EE is related to adipose tissue.It may act as an endocrine tissue releasing inflammatory cytokines and leptin, which may further exacerbate the esophageal inflammatory process 35 .
In terms of metabolic diseases other than obesity, DM may cause esophageal dysfunction, which results in the amplitude reduction of esophageal contractions, less peristaltic waves, decreased LESP (lower esophageal sphincter pressure), and abnormal gastroesophageal reflux 36,37 .This is consistent with our finding that the risk of EE is increased in diabetic patients.Interestingly, the esophageal dysfunction in diabetic patients is also associated with autonomic neuropathy involving the vagal nerve, especially when the patient is in hyperglycemic state or has diabetes for 5-10 years after onset 38,39 .Gastric emptying can be disrupted due to this process, which www.nature.com/scientificreports/triggers EE 39 .This process is further worsened by the fact that reflux symptoms may be more frequent in diabetic patients with three major complications (retinopathy, neuropathy, nephropathy) and longer duration of DM 40,41 .
In this study, we find that hypertension increases the risk of EE.This finding is first confirmed by Gudlaugsdottir et al. 42 , which finds a significantly higher systolic blood pressure (SBP) in EE compared to the controls, although the underlying pathophysiology is still unclear.The relationship between hypertension and esophageal reflux is further confirmed by Hu et al. 43 , which observes a significant improvement in the hypertension control after laparoscopic fundoplication during a 3.5 year follow-up period.
Our overall analysis finds dyslipidemia to be a risk factor for EE.However, most studies included in the analysis do not find dyslipidemia to be a risk factor.To evaluate this finding, we also separately analyzed several components of dyslipidemia, such as hypercholesterolemia, hypertriglyceridemia, high LDL-C, and low HDL-C.Our results suggest that hypertriglyceridemia is a risk factor of EE, but not dyslipidemia and its other components.Several studies have suggested triglyceride (TG) as an independent risk factor for EE related to humoral components that altered LESP and the frequency of transient relaxation 44,45 .TG has also been correlated with high fat intake, causing delayed gastric emptying time [46][47][48] .Moreover, hypertriglyceridemia is a significant predictive factor of EE severity, possibly related to fatty liver and insulin resistance 49 .The chronic inflammation in EE due to gastric acid injury may cause abnormal lipid metabolism, increasing TG 47 .Yet, several studies do not find TG to be an independent risk factor of EE 50,51 .
NAFLD also reaches statistical significance as a risk factor for EE.A study reports that only NAFLD is associated with EE, but not obesity 45 .NAFLD also increases the systemic oxidative stress and decreases the antioxidant capacity, which disrupts the gastric mucus layer and further causing esophageal mucosal damage and increasing the risk of EE 45 .
Interestingly, both gastric atrophy and H. pylori infection show to be protective factors for EE.The gastric atrophy can be classified into closed-type (C-type) and open type (O-type) according to the endoscopic atrophic border.According to Kim et al. 52 , the ambulatory pH monitoring study indicates that the O-type is associated with a lesser number of reflux symptoms and EE than the C-type.The O-type is characterized by an increasing number of impaired acid secreting parietal gastric cells will hinder more the gastric acid production, which will lead to hypochlorhydria, lessen the esophageal acidity, and further contribute to the pathogenesis of EE 52,53 .In a similar manner, the H. pylori infection may present protective mechanism since H. pylori chronic inflammation can cause gastric atrophy and further decreases the acid secretory capacity of the gastric lining 54,55 .It is only observable in O-type cases, while missing in the C-type, which produces higher gastrin and acid secretion 56 .However, this finding should be interpreted carefully since uneradicated H. pylori still carries a high risk of gastric cancer through several complex mechanisms 57 .Therefore, even though H. pylori is protective towards EE in our study, its eradication should still be well-considered to prevent the incidence of gastric cancer in later life.
To the best of our understanding, there has been no study that focuses on the meta-analysis of EE prevalence.We find that the prevalence of EE in America and Europe is higher than that in Asia.Recent meta-analyses on the prevalence of GERD 58 and BE 59 show similar results.A study by Qumseya et al. 60 also finds a higher pooled prevalence of BE in low-risk Western populations compared to non-Western populations.One explanation for this distribution may be the difference in lifestyles.The typical Western diet is known to be high in fat, sodium, calories, and sugar, while it is low in fiber, fruits and vegetables.Concurrently, we have identified that White/ Caucasian and individuals with obesity, type 2 diabetes mellitus, hypertension, dyslipidemia, and associated disease, such as NAFLD, are more significantly at risk of suffering from EE.Additionally, our meta-analysis shows a higher pooled prevalence of EE in Africa and the Middle East compared to those in other regions.This finding is in contrast to a previous BE meta-analysis by Eusebi et al. 59 , which finds the prevalence of BE in African and Middle Eastern countries to be lower than that in American countries.
We acknowledge several limitations in our study.First, we find some considerable high heterogeneities in most of the analyzed factors, mainly between the studies, such as population characteristics, various EE diagnostic criteria, differences in UGI study indications, and comorbidities along with various diagnostic criteria and cutoff values for their diagnosis.Second, although the EE diagnosis in the included studies is based on endoscopic result and the associated diagnostic criteria, endoscopy is still relatively an operator dependent-investigation, which may affect the EE prevalence in each country.Third, the number of included studies in several factors is still less than 10 studies; hence, the results should be carefully interpreted.Fourth, the included studies are mostly conducted in Asia (84 studies) and America (15 studies).This may affect the prevalence and risk factors of EE, and their interpretations in our study.Accordingly, we encourage more researchers from regions other than Asia to conduct more studies regarding the prevalence and risk factors of EE.However, regardless of the limitations, our study carries some strengths.The numbers of our included studies and their participants are relatively sufficient to cover a wide range of geographical areas; therefore, we can analyze the worldwide EE prevalence.
As the conclusion, we find several risk and protective factors of EE classified in three groups of factors, including demographical factors, comorbidities, and medication history.In the demographical factors, the risk of EE is increased due to age ≥ 60 y.o., being White/Caucasian, being single or unmarried, and having GERD ≥ 5 years.Interestingly, both risk and protective factors of EE are found in the comorbidities.Obesity, DM, hypertension, dyslipidemia, hypertriglyceridemia, HH, and NAFLD act as risk factors, while H. pylori infection and atrophic gastritis act as protective factors.The EE prevalence in each of America, Africa, and Europe are higher than that in Asia.Given these findings, an integrated care pathways of EE-including the decision regarding the timing of endoscopy based on the risk factors-is expected to be constructed, which then may help medical professionals to give proper and comprehensive managements for patients who are at a high risk of EE.

Methods
This systematic review and meta-analysis were conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) latest statement 61 .The protocol of this study has been previously registered to the International Prospective Register of Systematic Reviews (PROSPERO) database (CRD42023418716).

Search strategy.
A systematic computerized data searching of relevant studies was conducted in four electronic medical databases, including PubMed, Scopus, Cumulative Index to Nursing and Allied Health Literature (CINAHL) Plus database via EBSCOhost, and Web of Science, by two authors (A.P.W. and B.S.W.) from inception to June 2, 2022.The construction of keywords was performed based on Medical Subject Headings (MeSH) terms combined with their variance and other additional terms as following: "risk", "predict", "erosive esophagitis", "gastroesophageal reflux disease", and the variations of those terms.Boolean operators' combinations were also applied in order to broaden and narrow the search results.The search was restricted to human participants only with no language and publication date restrictions.

Eligibility criteria.
The relevant studies were included if they met several following inclusion criteria: (1) study design of observational study; (2) study participants consisted of adult patients aged 18 years or older who had undergone upper gastrointestinal (UGI) endoscopy, either to screen or to diagnose EE; and (3) the measured outcomes were odds ratios (ORs) of any possible risk factors related to EE and number of EE events.The exclusion criteria were as follows: (1) duplicate studies; (2) irrelevant titles and/or abstracts; (3) irretrievable full-texts; and (4) incorrect study design (review articles, clinical trials, systematic reviews, meta-analyses, case reports or series, letter to editors, conference abstracts).
Data extraction and quality assessment.All relevant studies were independently screened by seven of the co-authors.Any disagreements were resolved in a consensus involving all authors.The extracted data from the included studies were the author, year of publication, study location (country and region), study design, diagnostic guideline for EE, age, specific population characteristic, sample size, number of EE events, EE-related risk factors expressed in ORs, and the adjustment factors.We assessed the quality of the included studies using the Newcastle-Ottawa Scale (NOS) tool.For cohort and case-control studies, their quality was considered as good, moderate, or poor if their score was 7-9, 4-6, and 0-3, respectively.For cross-sectional studies, a score of 9-10 was considered as very good, 7-8 as good, 5-6 as satisfactory, and 0-4 as unsatisfactory.The quality assessment was conducted collaboratively through a group discussion by all authors, and the final decision was also taken based on the agreement of all authors.
Statistical analysis.Meta-analyses were performed for the outcome of pooled ORs in each EE-related risk factor using RevMan ver.5.4 (The Cochrane Collaboration, The Nordic Cochrane Centre, Copenhagen, Denmark).We also performed meta-analysis of pooled EE prevalence in each study using STATA ver.16.0 (Stata Corporation, College Station, TX, USA) as the secondary outcome.The heterogeneity among studies was assessed using chi-square test (Cochran's Q statistic).Then, we quantified the level of heterogeneity with the Higgins' I 2 statistic as follows: 0% was considered negligible heterogeneity, < 25% as low heterogeneity, 25-75% as moderate heterogeneity, and > 75% as high heterogeneity 62 .Since there was a considerable variability and diversity among studies and the characteristics of the study participants, we primarily applied the random-effect model (REM) for risk factors and prevalence analyses.P-value < 0.05 was considered statistically significant.The publication bias was visually assessed using funnel plot and quantitatively assessed using Egger's test.Sensitivity analysis was carried out using the leave-one-out method.

Figure 1 .
Figure 1.PRISMA flow diagram of the study selection process.