Prevalence of Barrett’s Esophagus and Adenocarcinoma With and Without Gastroesophageal Reflux: A Systematic Review and Meta-Analysis

Abstract

Background and Aims:

While gastroesophageal reflux disease (GERD) symptoms are an essential criterion for Barrett’s esophagus (BE) screening in most gastroenterology society guidelines, a significant proportion of BE and esophageal adenocarcinoma (EAC) cases do not endorse them. In a systematic review and meta-analysis (SRM), we aimed to study the prevalence of BE/EAC in those with and without GERD.

Methods:

A systematic search was conducted through 5 major databases for studies reporting prevalence of BE/EAC in patients with and without GERD. Pooled proportions and odds ratios (OR) of BE, long-segment BE (LSBE), short-segment BE (SSBE), dysplasia and EAC in patients with and without GERD were synthesized.

Results:

Forty-three articles (12,883 patients with; 51,350 patients without GERD) were included in the final analysis. BE prevalence was 7% (95% CI 5.8–8.5) and 2.2% (1.6–3) among individuals with and without GERD, respectively. EAC prevalence was 0.6% (0.4–1) and 0.1% (0–0.2) in those with and without GERD, respectively. The overall risks for BE (OR=2.91; 2.06–4.11) and LSBE (OR=4.17; 1.78–9.77) were higher in patients with GERD, but the risk for SSBE (OR=1.77; 0.89–3.52) did not differ between the two groups. In 9 population-based high-quality studies (2244 patients with; 3724 patients without GERD), BE prevalence in patients without GERD was 4.9% (95% CI 2.6–9). BE prevalence was highest in North American studies (10.6% [GERD] and 4.8% [non-GERD]).

Conclusion:

BE prevalence in those without GERD is substantial, particularly in large high-quality population-based studies. These data are important to factor in future BE/EAC early detection guidelines.

Graphical Abstract

Introduction

Esophageal adenocarcinoma (EAC) is an aggressive malignancy with 20% survival rate at 5 years in Western populations.¹ Survival with early stage EAC is substantially higher emphasizing the importance of early detection and cancer prevention.² Barrett’s esophagus (BE) is currently the only known precursor to EAC increasing the risk by 30-to-125 fold.^3,4 BE is characterized by intestinal metaplasia of the distal esophagus, specifically the conversion of squamous epithelium to columnar epithelium with goblet cells in response to injury and aberrant healing in the setting of gastroesophageal reflux (GERD).³ Despite BE screening recommendations and efforts, EAC incidence has increased sixfold over the last 4 decades.^3,5

Currently, all major gastroenterology societies recommend screening for BE in patients with risk factors which include chronic reflux symptoms, male sex, age > 50 years old, Caucasian race, history of tobacco smoking, central obesity, and family history of BE/EAC in first-degree relatives.^3,5–7 An essential criterion for BE screening based on most gastroenterology guidelines is the presence of GERD, focusing on symptom frequency and duration, due to the fact that numerous studies have long identified a significant association of BE and EAC with GERD symptoms.^8–10 GERD is associated with a threefold increased risk for BE.¹¹ Moreover, the association of GERD and long-segment BE (LSBE), defined as a Barrett’s segment length ≥3 cm above the gastroesophageal (GE) junction, is more robust compared to short-segment BE (SSBE), defined as a Barrett’s segment length <3 cm above the GE junction.^3,5,11 However, several studies have reported a surprisingly high prevalence of BE in patients who do not report chronic reflux symptoms (15–27%).^3,12–14 Additionally, a substantial proportion of EAC patients also do not report a significant history of GERD symptoms.^3,5 Whether this is due to the known hyposensitivity to the perception of reflux or the independent effects of other risk factors such as obesity and smoking on BE and EAC risk remains to be determined.

Indeed, it has been shown that current guidelines requiring the presence of reflux symptoms for BE screening have low sensitivity and only moderate specificity,^3,4,15 and support for considering BE screening in those without GERD but with other risk factors has grown.^3,16 With the advent of minimally invasive non-endoscopic tools for BE detection, this issue takes on additional significance.^17–19 Specifically, swallowable cell-collection devices such as Cytosponge, EsoCheck and EsophaCap, combined with BE specific biomarkers such as methylated DNA markers and trefoil factor 3 (TTF3),^17–19 are now guideline supported as alternatives to sedated endoscopy.¹⁶ Given the invasive and expensive nature of sedated endoscopy, screening recommendations were initially limited to those with chronic reflux, acknowledging that BE and EAC in those without reflux symptoms will likely be missed. However, with the likely lower cost of non-endoscopic BE detection tools, the possibility of wider screening in those without GERD but with other risk factors has arisen.

Taken together, given these issues and the importance of accurately defining the prevalence of BE in those without reflux, in this systematic review and meta-analysis (SRM), we aimed to determine the prevalence of BE/EAC in those without GERD and compare this with the prevalence of BE/EAC in those with GERD. GERD was defined as the presence of one or more of the following: heartburn, acid regurgitation, symptoms compatible with GERD according to questionnaires, medication use for GERD, or endoscopic evidence of erosive esophagitis. We hypothesized that there will be no prominent difference in the prevalence of and/or risk for BE/EAC in those with and without GERD, and especially for SSBE.

Methods

The study protocol was registered to PROSPERO (CRD42022380529). This systematic review and meta-analysis was conducted and reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.²⁰ The meta-analysis was performed in accordance with the Meta-analysis of observational studies in epidemiology (MOOSE) checklist.²¹

Search Strategy and Study Identification

An experienced medical librarian (DG) conducted a comprehensive systematic literature search through 5 major databases including MEDLINE, EMBASE, Scopus, Cochrane CENTRAL, and Web of Science since inception to 7/29/2022. Search strategies and key words used in the literature search are summarized and provided in supplementary materials (Appendix A). The search was restricted to studies in English of human subjects that were published in peer-reviewed journals.

Study Selection

De-duplication and screening of articles were undertaken using EndNote reference management software (Clarivate 2022, EndNote X9) and systematic review tool, Covidence.^22,23 Specifically, two authors (BS and KP) independently screened all titles and abstracts of candidate studies identified in the literature search for relevance and excluded studies that did not address the research question of interest. The remaining articles underwent a full-text assessment to determine eligibility for qualitative and/or quantitative data synthesis based on the inclusion and exclusion criteria summarized in Appendix B. Briefly, the inclusion criteria involved randomized controlled trials (RCT), prospective/retrospective cohort studies, and cross-sectional investigations irrespective of study geography in English that reported explicitly or implicitly (via ability to calculate) the prevalence of BE/EAC in both the GERD and non-GERD groups. Specifically, BE must have been diagnosed with endoscopy AND pathological confirmation. GERD could have been defined/diagnosed by symptoms, validated reflux symptom questionnaires, medication (such as proton-pump inhibitors) use for GERD, endoscopic assessment revealing esophagitis or by ambulatory pH monitoring. Case-control studies were excluded as the control arm would have no BE/EAC patients whereas the case arm would have all BE/EAC patients making the overall data unusable. Any discrepancy in article selection was resolved by discussion or adjudicated by the senior author (PGI). Additional studies identified through a manual search were also included if the inclusion criteria were met.

Data Abstraction

All data was independently extracted by BS and KV. Specifically, from the eligible studies, information regarding study identification (first author’s last name, year of publication), study settings (single- or multi-center or population-based study, outpatient/endoscopy clinic/inpatient, country of enrollment, dates of study enrollment), study design (RCT, prospective cohort, retrospective cohort, cross-sectional), sample size (initial sample size, final sample size), population demographics (mean age, number of males and females, race of patients, smoking prevalence, prevalence of obesity, average BMI), GERD and BE/EAC details (GERD frequency/duration, how GERD was defined/diagnosed, number of LSBE and SSBE, Prague criteria utilization,²⁴ presence/severity of dysplasia, EAC stage), primary outcomes (prevalence or calculated prevalence of BE or EAC in those without a history of and/or current GERD, prevalence or calculated prevalence of BE or EAC in those with a history of or current GERD, associated crude and/or adjusted odd ratios (OR) with p-values and/or 95% confidence intervals), details regarding the prevalence of other risk factors for BE (number of patients with BE of age >50 years old, male sex, Caucasian race, hiatal hernia diagnosis, central obesity/obesity, family history of BE and/or EAC in a first-degree relative, and/or history of cigarette smoking), as well as information required for quality assessment was collected using a standardized data collection form.

Quality Assessment

Assessment of methodologic quality for cohort and cross-sectional studies was determined using a modified version of the Newcastle-Ottawa scale (NOS), as provided in Appendix C, to classify studies as high-quality (low risk-of-bias), medium/acceptable-quality (average risk-of-bias) and low-quality (high risk-of-bias).²⁵ Additionally, assessment of methodologic quality for RCTs was determined using the Scottish Intercollegiate Guidelines Network (SIGN) quality checklist for RCTs.²⁶ Both tools were utilized independently by BS and KP.

Qualitative Parameters Assessed

Qualitative data synthesis, in addition to summarizing descriptive data, focused on determining the number of studies that reported a significant difference in BE/EAC prevalence between those with and without GERD compared to the number studies that did not identify such a difference. The percentage of BE patients that were >50 years old, male, Caucasian, with a hiatal hernia diagnosis, central obesity/obesity, family history of BE and/or EAC in a first-degree relative, and/or history of cigarette smoking was also assessed.

Outcomes Assessed

Primary outcomes of interest were pooled proportions of BE, LSBE, SSBE, dysplasia and EAC prevalence in patients with and without GERD assessed by symptoms, questionnaires, medication use or evidence of erosive esophagitis. The risk for these conditions in patients without versus with GERD were assessed by calculating pooled unadjusted ORs. We performed various pre-planned subgroup analyses: limited to population-based screening studies, study risk-of-bias, study geography (North America, Asia, Europe), and method of GERD diagnosis (questionnaire, symptoms). We hypothesized that there will be no prominent difference in the prevalence of BE in those with and without GERD when analysis is limited to large population-based screening studies or when focusing solely on studies deemed low/average risk of bias. Additionally, we hypothesized that North American studies will demonstrate the highest BE prevalence regardless of GERD status given higher prevalence of other established BE risk factors such as obesity in this region.

Statistical Analysis:

Descriptive statistics were calculated using Microsoft Excel (Microsoft Corporation 2022, Microsoft Excel) to summarize study characteristics. Specifically, means and standard deviations were calculated for continuous variables whereas percentages and frequencies were calculated for categorical variables unless stated otherwise. For quantitative data synthesis, including pooled prevalence data and unadjusted ORs as well as sub-group analyses, standard meta-analysis methods were followed using random-effects model as suggested by DerSimonian and Laird.²⁷ The application can be seen to fit within their general approach where effect is measured by probability of risk. Heterogeneity was assessed using Cochrane Q statistics and I2% statistics. Values of <30%, 30% to 60%, 61% to 75% and >75% were considered suggestive of low, moderate, substantial, and considerable heterogeneity. Publication bias was assessed qualitatively by visual inspection of funnel plot and quantitatively by Egger’s test. Meta-analysis synthesis was performed using Comprehensive Meta-Analysis (CMA) software, version 4 (BioStat, Englewood, NJ).

Results

Study Identification and Selection

Our search yielded a total of 16,345 citations, of which 121 were eligible for full-text assessment after de-duplication, and title and abstract screening (Figure. 1). One study of interest was also found through the manual search and was eligible for full-text assessment. Of the remaining articles, 43 met the inclusion criteria and were utilized for data synthesis (Figure. 1).^{4,5,12–14,28–65}

Fig. 1.

PRISMA Flow Diagram Summarizing Search and Review Strategy.

Study Characteristics and Participant Demographics

Of the 43 studies, 39 were published in the year 2000 or later (Table 1). A large proportion of these studies were conducted in North America (17/43), followed by Asia (16/43), Europe (8/43) and South America (2/43). While the majority of the studies were performed at a single-center (27/43), there were 7 multi-center investigations and 9 large population-based studies (Table 1). Most of the studies employed a cohort study design (30/43), followed by 11 cross-sectional investigations and 2 RCTs. Furthermore, the median (range) study duration was 22 (3–139) months with a median (range) final sample size of 961 (87–14898) participants. The mean average (SD) age of these participants reported in the studies was 55.5 ± 6.8 years, and the overall proportion of females was 47.4%.

Table 1.

Study Characteristics and Participant Demographics

Study Characteristics
Publication Year in or after 2000, N(%)	39 (90.7)
Study Duration, Median (Range) in Months	22 (3–139)
Sample Size, Median (Range)	961 (87–14898)
Study Design
Cohort, N(%)	30 (69.8)
Cross-sectional, N(%)	11 (25.6)
Randomized Controlled Trial, N(%)	2 (4.6)
Study Setting
Single-center, N(%)	27 (62.8)
Multi-center, N(%)	7 (16.3)
Large Population-based, N(%)	9 (20.9)
Participant Demographics
Average Age Reported in Studies, Mean (SD)	55.5 (6.8)
Sex (% Female)	47.4

GERD and BE Characteristics

Amongst the 43 studies, GERD was predominantly defined through the use of reflux symptom questionnaires (48.8%) or the presence of symptoms (46.5%). Only one study utilized medication use, and another used endoscopic assessment revealing esophagitis, as criteria to define the presence of GERD. Of the forty-three studies, only 9 utilized the Prague criteria when describing BE. Qualitatively, 11/43 studies did not find a significant difference in the prevalence of BE in those with and without GERD, whereas 13 studies did report a difference. The remaining studies (19/43) did not provide this information explicitly. Only some studies reported crude (6) and adjusted (7) ORs comparing BE prevalence in those with and without GERD. Of note, in the 3/7 studies that reported an adjusted OR but did not find a significant difference in the BE prevalence in those with and without GERD, all 3 adjusted for waist-to-hip ratio/BMI. Supplementary Table 1 summarizes the adjusted ORs and the variables that were adjusted.

Prevalence of Risk Factors for BE

In assessing risk factors for BE other than GERD, in the 12 studies that provided enough information regarding age, the percentage of BE patients (irrespective of GERD status) that were >50 years old was 84.8% (Supplementary Table 2). Similarly, with respect to sex and race, the percentage of BE patients that were male or Caucasian was 67.7% and 51.3%, respectively. Lastly, the percentages of BE patients that had a hiatal hernia, central obesity/obesity, or a history of smoking were 49.7%, 39.9% and 44.0%, respectively. No studies provided enough information on family history of BE and/or EAC in a first-degree relative to investigate this risk factor further. While we attempted to investigate the prevalence of risk factors for BE in the non-GERD group alone, only 2 studies provided relevant patient level data, preventing quantitative data synthesis. Fan et al. reported the prevalence of male participants with BE in the non-GERD group to be 66.1% compared to 72.2% in the GERD group.³⁸ The second study reported the prevalence of age > 50 years, male sex, hiatal hernia, BMI > 25 kg/m², and smoking history in BE patients without GERD, to be 18.2%, 63.4%, 0%, 27.3%, and 9.1%, respectively.⁵⁰ In comparison, the prevalence of these risk factors in the GERD group was 75%, 43.8%, 18.8%, 25% and 12.5%, respectively. Similarly, in the 3 studies that provided adjusted ORs, and did not find a significant difference in the BE prevalence in those with and without GERD, all adjusted for waist-to-hip ratio/BMI. This may suggest that central obesity/obesity was an independent risk factor for BE in these studies.

Quality (Risk of Bias) Assessment

Following the pre-defined criteria outlined in the methodology, overall, our systemic review included predominately average risk-for-bias (27/43) and low risk-for-bias (8/43) studies, with the remaining articles being classified as high risk-for-bias (8 studies). The detailed risk-of-bias assessment is summarized in Supplementary Table 3.

Meta-analysis Outcomes

The forty-three articles encompassed 12,883 participants with GERD and 51,350 participants without GERD who were included in the final analysis (Table 2). Overall, the prevalence of histologically confirmed BE was 7% ([95% CI 5.8–8.5; I²=89%]) and 2.2% [95% CI 1.6–3; I²=95%] among individuals with and without GERD symptoms, respectively. Similarly, the prevalence of LSBE and SSBE were also higher in the GERD group (Table 2). For both the GERD and non-GERD cohorts, the prevalence of SSBE (8.5% and 6.8%, respectively), was greater than LSBE (2.8% and 0.9%, respectively). The overall risk for BE was significantly higher in patients with GERD symptoms (OR=2.91; [95% CI 2.06–4.11; I²=84%]) as was the risk for LSBE (OR=4.17; [95% CI 1.78–9.77; I²=0%]). However, there was no significant difference in the risk for SSBE between the two groups (OR=1.77; [95% CI 0.89–3.52; I²=70%]). Importantly, the prevalence of dysplasia was 0.6% ([95% CI 0.3–1.1; I²=7%]) and 0.2% ([95% CI 0.1–0.6; I²=74%]) in those with and without GERD but the risk for dysplasia was not different between the two groups (OR=1.55; [95% CI 0.52–4.58; I²=0%]). While the prevalence of EAC was 0.6% [95% CI 0.4–1.0; I²=3%] and 0.1% [95% CI 0–0.2; I²=26%] among individuals with and without GERD symptoms, respectively, similar to dysplasia, there was no statistically significant difference in the risk for EAC between the two groups (OR=5.19; [95% CI 0.69–39.22; I²=0%]). This is further highlighted in the forest plots outlined in Figure 2.

Table 2.

Summary of Pooled Rates of Barrett’s Esophagus and Esophageal Adenocarcinoma Prevalence in those with and without Gastroesophageal Reflux Disease and corresponding Un-adjusted Odds Ratios

Outcomes	Pooled rate (95% confidence interval; I2%)
	Cohort with GERD	Cohort without GERD
BE	7% (5.8–8.5; 89%); [43 studies, 12883 pts]	2.2% (1.6–3; 95%); [43 studies, 51350 pts]
Symptoms	7.7% (5.7–10.5; 90%); 20 studies	1.6% (0.9–3; 96%); 20 studies
Questionnaire	7.2% (5.1–10; 87%); 21 studies	2.8% (1.7–4.7; 95%); 21 studies
LSBE	2.8% (1.4–5.4; 70%); [10 studies, 2211 pts]	0.9% (0.4–2.1; 63%); [9 studies, 1726 pts]
Symptoms	1.6% (0.2–13.8; 74%); 3 studies	0.8% (0.3–2.6; 72%); 3 studies
Questionnaire	2.9% (1.4–5.9; 45%); 7 studies	0.9% (0.2–3.5; 0%); 6 studies
SSBE	8.5% (4.5–15.5; 92%); [10 studies, 2211 pts]	6.8% (3.4–13.2; 90%); [9 studies, 1726 pts]
Symptoms	13.7% (3.6–40.2; 97%); 3 studies	10.5% (3.8–26; 90%); 3 studies
Questionnaire	7.3% (3.5–14.7; 79%); 7 studies	4.7% (1.8–11.7; 90%); 6 studies
BE with dysplasia	0.6% (0.3–1.1; 7%); [10 studies, 2673 pts]	0.2% (0.1–0.6; 74%); [10 studies, 12525 pts]
EAC	0.6% (0.4–1; 3%); [17 studies, 5256 pts]	0.1% (0–0.2; 26%); [17 studies, 29264 pts]
Un-adjusted Odds Ratio		P-value
BE	2.91 (2.06–4.11; 84%); 43 studies	<0.01
LSBE	4.17 (1.78–9.77; 0%); 9 studies	0.001
SSBE	1.77 (0.89–3.52; 70%); 9 studies	0.1
BE with dysplasia	1.55 (0.52–4.58; 0%); 10 studies	0.4
EAC	5.19 (0.69–39.22; 0%); 17 studies	0.1

BE: Barrett’s esophagus; LSBE: Long segment Barrett’s esophagus; SSBE: Short segment Barrett’s esophagus; EAC: Esophageal adenocarcinoma; GERD: Gastro-esophageal reflux disease

Fig. 2.

Forest plots demonstrating unadjusted odds ratios (OR) for Barrett’s esophagus (BE; A), long segment BE (LSBE; B), short segment BE (SSBE; C), BE with dysplasia (D), and Esophageal adenocarcinoma (EAC; E) in those with and without gastroesophageal reflux disease (GERD).

A Priori Determined Subgroup Analyses

On subgroup analysis limited to large high quality population-based screening studies (9 studies; 2244 patients with and 3724 patients without GERD) at low risk of bias,^{4,5,12,13,53–55,59,65} the prevalence of BE in patients without GERD was 4.9% [95% CI 2.6–9; I²=93%], compared to 7.7% in BE patients with GERD [95% CI 4.4–12.9; I²=90%; Table 3]. Similarly, the overall risk for BE was higher in patients with GERD symptoms (OR=1.9; [95% CI 1.5–2.4; I²=0%]). Subgroup analyses based on study risk-of-bias (high, average and low) and by study geography are summarized in Table 4. For both those with and without GERD, the BE prevalence was found to be higher in studies deemed to be at low/average risk-of-bias when compared to studies rated as high risk-of-bias. For both the GERD and non-GERD cohorts, BE prevalence was the highest amongst studies from North America (10.6% and 4.8%, respectively), followed by those from Europe (8.6% and 2.1%, respectively), and those from Asia (3.5% and 1%, respectively).

Table 3.

Summary of pooled outcomes and Un-adjusted Odds Ratio of Large Population-based Studies with Low Possibility of Bias

Outcomes	Pooled rate (95% confidence interval; I2%)
	Cohort with GERD	Cohort without GERD
BE	7.7% (4.4–12.9; 90%); [9 studies, 2244 pts]	4.9% (2.6–9; 93%); [9 studies, 3724 pts]
LSBE	4.9% (1.5–15.1; 85%); 4 studies	NA (only 2 studies)
SSBE	3.7% (0.7–18.2; 87%); 3 studies	NA
BE with dysplasia	NA (only 2 studies)	NA
EAC	NA	NA
	Un-adjusted odd ratio	P-value
BE	1.9 (1.5–2.4; 0%); 9 studies	<0.01

BE: Barrett’s esophagus; LSBE: Long segment Barrett’s esophagus; SSBE: Short segment Barrett’s esophagus; EAC: Esophageal adenocarcinoma; GERD: Gastro-esophageal reflux disease

Sensitivity Analysis and Heterogeneity

To assess whether any individual study had a dominant effect on the meta-analysis, we excluded one study at a time and analyzed its effect on the major findings. In this analysis, no single study significantly affected the outcome or the heterogeneity. Given that one study in our investigation utilized reflux esophagitis as the diagnostic criterion for GERD, we excluded this study as part of the sensitivity analysis.⁵⁹ In the remaining 42 studies, prevalence of BE was 7.5% [95% CI 6.1–9.3] and 2.2% [95% CI 1.5–3.3] among individuals with and without GERD symptoms, respectively. The overall risk for BE was significantly higher in patients with GERD symptoms (OR=2.91; [95% CI 2.05–4.15]). Similarly, in the sub-group analysis of the remaining 8 high quality large population-based studies, the prevalence of BE in patients without GERD was 5.7% [95% CI 3.1–10.4], compared to 8.4% [95% CI 4.8–14.4] in patients with GERD; the risk for BE was greater in the GERD group (OR=1.80; [95% CI 1.39–2.31]). While this study provided information on esophagitis grading, it did not report the specific grades for the BE patients.

Overall, there was considerable heterogeneity in the comparison of overall BE prevalence between those with and without GERD. However, the comparisons of prevalence of LSBE (questionnaire), BE with dysplasia and EAC exhibited lower heterogeneity.

Discussion

In this SRM of 43 studies including 64,233 participants, the overall prevalence of BE among individuals with GERD was 7%, compared to 2.2% in those without GERD. The prevalence of EAC in those with and without GERD was 0.6% and 0.1% respectively. However, on subgroup-analysis of large high-quality population-based screening studies with lower risk of bias, the BE prevalence in those without GERD was found to be more substantial at 4.9%. This prevalence was similar to that seen in all studies from North America. Notably, the prevalence of BE with dysplasia was comparable between patients with and without GERD.

While GERD was associated with an increased BE risk when analyzing data from all studies, the risk of BE with dysplasia was comparable in patients without GERD (OR=1.55; [95% CI 0.52–4.58; I²=0%]; p = 0.4).^{4,30–32,35,36,38,51,53,58} This finding may have important implications given that dysplasia precedes most EACs.⁶⁶ That is, it may suggest that patients without GERD symptoms, who harbor BE, but are currently not targeted for screening by most guidelines, may progress to dysplasia at a similar rate to those with GERD, who are screened. This may result in the delayed diagnosis of dysplasia leading to missed opportunities for the prevention and treatment of EAC.^3,5 While the risk for EAC did not reach statistical significance between the two groups, this may reflect type 2 error due to a smaller number of patients.

On subgroup-analysis of only population-based screening studies of higher quality, at lower risk of bias, we found that BE prevalence in those with GERD was 7.7% (similar to that found in the 43 studies overall). However, notably BE prevalence was 4.9% in those without GERD. This suggests that based on the majority of current gastroenterology guidelines, which limit targeted BE screening to only patients with GERD, we may be missing ~39% of BE cases, substantially reducing their sensitivity, and consequently, their effectiveness.^3,4,15 This may be particularly applicable to North America, where the prevalence of BE in those without GERD was highest, at 4.8% (similar to that of population-based studies). Studies at reduced risk-of-bias also reported higher BE prevalence in those without GERD compared to studies deemed to be at high risk-of-bias (Table 4), suggesting that lower quality studies may be underestimating BE prevalence in those without GERD.

BE prevalence was highest amongst studies in North America, followed by Europe and then Asia regardless of GERD status. Indeed, the BE prevalence was approximately 2.3 and 4.8 times greater in North America when compared to Europe and Asia, in those without GERD, respectively. The prevalence of BE in Asia in our study is similar to what is reported in the literature.⁶⁷ Some BE risk factors (such as GERD and obesity) are less frequent in Asia than in North America, which may explain the overall lower BE prevalence in this region.

Expanding BE screening with EGD to those without reflux symptoms (but with other risk factors), has been contemplated but is not without substantial implications.³ Chandar et al. investigated the population eligible for BE screening in the US and found that it is large and variable based on the different gastroenterology society guidelines.⁶⁸ The screen-eligible population ranged from 19.7–51.9 million (using GERD as an essential criterion, as in the ACG, ASGE and BSG guidelines) to 120.1 million (using GERD as only one of the risk factors), as suggested in the 2022 AGA Clinical Practice Update.⁶⁸ Use of sedated endoscopy, to screen such a large number of individuals is likely impractical, given costs and current limitations in access. Conversely, if higher risk individuals in the population without GERD symptoms, but with other BE risk factors, could be identified, this may allow prioritization and targeted screening in this population.⁶⁸ Non-endoscopic BE screening technologies which are now guideline supported may be reasonable alternatives.¹⁶ Swallowable cell-collection devices with biomarkers were found to be cost effective screening tests on Markov modeling in individuals without GERD, with a BE prevalence of 5%, which is similar to the prevalence we report in the subgroup analysis limited to population based studies.⁶⁹

Given that the prevalence of BE is ~5% in those without GERD, the second major implication of this finding is that other risk factors likely contribute to the risk for BE. In this study, we found that the prevalence of known risk factors for BE, including age ≥ 50, male sex, Caucasian race, central obesity/obesity, smoking history, and presence of hiatal hernia was high, with age ≥ 50 and male sex being the most prevalent. Limited data from two studies seem to suggest that some of the conventional risk factors for BE such as older age, male sex and obesity may also be risk factors of BE in those without GERD. While there is evidence that increasing number of risk factors in those with GERD is associated with higher BE prevalence³, larger studies to define the risk factors of BE in those without GERD are critical and warranted.

Our study had numerous strengths. First, we performed a comprehensive systematic search of the literature utilizing multiple databases with the assistance of an expert medical librarian. We also utilized various standardized tools for assessing study quality (NOS, SIGN checklist), for performing the meta-analysis and reporting our findings (PRISMA Guidelines), and we registered our protocol to PROSPERO. Additionally, our investigation had rigorous evaluation of study quality as we assessed study risk-of-bias in two different ways and performed sub-group analyses for both. We also performed various subgroup and sensitivity analyses to assess the consistency of the findings and to identify factors that may be associated with heterogeneity. Lastly, we excluded studies or data from ultra-short segment (< 1 cm) BE when explicitly described in studies.

We also acknowledge some limitations in our study. There was substantial heterogeneity in some of our analyses. This is likely due to the variability in the methods of GERD diagnosis, study designs and geographical locations. However, estimates of LSBE and EAC prevalence had low heterogeneity. Additionally, given inconsistent reporting and/or definitions between studies, we were unable to characterize GERD duration/frequency, impairing our ability to perform sub-group analyses of these variables. Similarly, the majority of studies did not provide relevant information on proton-pump inhibitor (PPI) use. In some studies, when PPI use was described, interestingly, this data was used to supplement the GERD diagnosis made by symptoms/questionnaire, while others used recent PPI exposure as an exclusion criterion due to concerns of masking of symptoms. Studies reporting PPI use in those with and without GERD symptoms, and its relationship to BE prevalence were scarce. Lastly, we were unable to account for collinear risk factors that increase the risk for both GERD and BE, such as obesity, as we did not have access to individual patient level data.

In summary, we conducted the first SRM on BE/EAC prevalence in those with and without GERD. While the overall prevalence of BE, especially LSBE, was greater in those with compared to those without GERD, in large population-based screening studies at lower risk-of-bias, the prevalence of BE was ~5% in those without GERD. BE in those without GERD may share some risk factors with those with GERD. Studies assessing BE prevalence and risk factors in those without GERD are needed. Additionally, the generation of a risk prediction model for BE in those without reflux symptoms may allow targeted screening in this population, particularly with emerging non-endoscopic technologies.

Table 4.

Subgroup Analyses of Barrett’s Esophagus Prevalence based on Study Risk of Bias and Geography

	Pooled rate (95% confidence interval; I2%)
	Cohort with GERD	Cohort without GERD
Study ‘risk of bias’:
A) Low	7.1% (4.3–11.7; 85%); 8 studies	2% (0.8–4.7; 96%); 8 studies
B) Average	8.1% (6.2–10.4; 90%); 27 studies	2.6% (1.6–4.2; 96%); 27 studies
C) High	5.3% (2.5–10.8; 86%); 8 studies	1.3% (0.6–3.2; 83%); 8 studies
Study Geography:
A) Asia	3.5% (2.3–5.2; 84%); [16 studies, 5625 pts]	1% (0.7–1.5; 91%); [16 studies, 25192 pts]
B) Europe	8.6% (5–14.4; 93%); [8 studies, 3039 pts]	2.1% (0.6–7.1; 98%); [8 studies, 18006 pts]
C) N. America	10.6% (8.2–13.5; 80%); [17 studies, 3791 pts]	4.8% (2.9–7.9; 93%); [17 studies, 7801 pts]
D) S. America	NA (only 2 studies)	NA (only 2 studies)

GERD: Gastro-esophageal reflux disease

Data Transparency Statement:

Data not in the article or supplementary material are available upon reasonable request to the corresponding author.

What You Need to Know:

BACKGROUND:

Gastroesophageal reflux disease (GERD) symptoms are an essential criterion for Barrett’s esophagus (BE) screening, but are not reported by a significant proportion of BE and esophageal adenocarcinoma (EAC) cases. This leads to reduced sensitivity and effectiveness of current guidelines. It is critical to assess the prevalence of BE and EAC in those without GERD symptoms.

FINDINGS:

The overall prevalence of BE was 7% and 2.2% among individuals with and without GERD, respectively. In 9 population-based studies with low risk of bias, BE prevalence in those without GERD was 4.9%.

IMPLICATIONS FOR PATIENT CARE:

BE prevalence in those without GERD is substantial and important to factor in BE/EAC early detection guidelines. Generation of risk prediction models for BE in those without GERD is warranted.

Appendix A: Summary of Search Strategy

Search strategies were created using a combination of keywords and standardized index terms. Searches were run on July 29, 2022 in Ovid Cochrane Central Register of Controlled Trials (1991+), Ovid Embase (1974+), Ovid Medline (1946+ including epub ahead of print, in-process & other non-indexed citations), Scopus (1788+), and Web of Science Core Collection (Science Citation Index Expanded 1975+ & Emerging Sources Citation Index 2015+).

Cochrane Central Register of Controlled Trials (CCTR) via Ovid (1991+)):

#	Query	Results from 29 Jul2022
1	(barrett* or ((columnar or metaplasia or adenocarcinoma or cancer* or carcinoma) adj3 ?esophag*) or endobrachyoesophagus).ab,hw,ti.	5,711
2	(prevalence* or incidence or morbidit* or comorbid* or epidemiolog* or distribution or population* or cohort* or longitudinal or follow-up or followup or prospective or retrospective or control* or cross-section* or cross-over or random* or trial* or matched-pair* or sampl* or multicenter or study or surveillance or biosurveillance or community-assessment* or common* or frequen* or rate* or ((single or doubl* or tripl* or treb*) adj3 (blind* or mask*))).ab,ti,hw,fx.	1,736,358
3	(reflux or regurgitation or G?RD or heartburn).ab,ti,hw.	11,992
4	1 and 2 and 3	355
5	limit 4 to english language	335
6	(Journal: Conference Abstract or Conference Abstract).pt.	207,233
7	(case-control or case-report or review).pt,ti.	15,531
8	(animal-model* or mouse or mice or murine or rat or rats or rodent* or porcine or swine or pig* or canine* or dog* or baboon*).ti,ab,hw.	30,977
9	(child* or p?ediatric* or infant* or newborn* or neonat* or girl* or boy* or toddler* or adoles* or teen* or youth).ti.	148,773
10	or/6–9	379,075
11	5 not 10	262

Embase via Ovid (1974+):

#	Query	Results from 29 Jul2022
1	exp Barrett esophagus/ or esophageal adenocarcinoma/	27,080
2	(barrett* or ((columnar or metaplasia or adenocarcinoma or cancer* or carcinoma) adj3 ?esophag*) or endobrachyoesophagus).ab,kf,ti,dq.	79,632
3	or/1–2	85,657
4	exp epidemiological data/ or exp epidemiology/ or clinical study/ or exp case control study/ or exp clinical trial/ or community trial/ or exp longitudinal study/ or major clinical study/ or open study/ or prospective study/ or retrospective study/	9,547,477
5	(prevalence* or incidence or morbidit* or comorbid* or epidemiolog* or distribution or population* or cohort* or longitudinal or follow-up or followup or prospective or retrospective or control* or cross-section* or cross-over or random* or trial* or matched-pair* or sampl* or multicenter or surveillance or biosurveillance or community-assessment* or common* or frequen* or rate* or ((single or doubl* or tripl* or treb*) adj3 (blind* or mask*))).ab,kf,ti,hw,fx,dq.	23,068,587
6	or/4–5	23,848,147
7	exp gastroesophageal reflux/	71,319
8	(reflux or regurgitation or G?RD).ab,kf,ti,hw,dq.	235,383
9	or/7–8	235,383
10	3 and 6 and 9	9,216
11	limit 10 to english language	8,477
12	limit 11 to conference abstract	2,432
13	(case-control or case-report or review).pt,ti.	3,617,637
14	(exp animal/ or animal experiment/ or nonhuman/) not (exp human/ or human experiment/)	6,952,861
15	exp juvenile/ not exp adult/	2,429,093
16	or/12–15	12,376,896
17	11 not 16	4,100

MEDLINE via Ovid (1946+ and Epub Ahead of Print, In-Process & Other Non-Indexed Citations and Ovid MEDLINE(R) Daily):

#	Query	Results from 29 Jul2022
1	Barrett Esophagus/	8,523
2	(barrett* or ((columnar or metaplasia or adenocarcinoma or cancer* or carcinoma) adj3 ?esophag*) or endobrachyoesophagus).ab,kf,ti.	54,866
3	or/1–2	55,614
4	exp morbidity/ or exp Epidemiologic Methods/	7,132,611
5	(prevalence* or incidence or morbidit* or comorbid* or epidemiolog* or distribution or population* or cohort* or longitudinal or follow-up or followup or prospective or retrospective or control* or cross-section* or cross-over or random* or trial* or matched-pair* or sampl* or multicenter or study or surveillance or biosurveillance or community- assessment* or common* or frequen* or rate* or ((single or doubl* or tripl* or treb*) adj3 (blind* or mask*))).ab,kf,ti,hw,fx.	19,238,137
6	or/4–5	19,980,055
7	exp Gastroesophageal Reflux/	28,543
8	(reflux or regurgitation or G?RD or heartburn).ab,kf,ti,hw.	113,748
9	or/7–8	113,826
10	3 and 6 and 9	4,774
11	limit 10 to english language	4,264
12	(case-control or case-report or review).pt,ti.	3,531,312
13	exp Animals/ not Humans/	5,032,606
14	(exp infant/ or exp child/ or adolescent/) not exp adult/	2,068,534
15	or/12–14	10,234,875
16	11 not 15	2,813

Scopus via Elsevier (1788+):

( ( ( TITLE-ABS-KEY ( barrett* OR endobrachyoesophagus ) OR TITLE-ABS-KEY ( ( columnar OR metaplasia OR adenocarcinoma OR cancer* OR carcinoma ) W/3 esophag* ) ) ) AND ( ( TITLE-ABS-KEY ( prevalence* OR incidence OR morbidit* OR comorbid* OR epidemiolog* OR distribution OR population* OR cohor t* OR longitudinal OR follow-up OR followup OR prospective OR retrospective OR control* OR cross-section* OR cross-over OR random* OR trial* OR matched-pair* OR sampl* OR multicenter OR study OR surveillance OR biosurveillance OR community-assessment* OR common* OR frequen* OR rate* ) OR TITLE-ABS-KEY ( ( single OR doubl* OR tripl* OR treb* ) W/3 ( blind* OR mask* ) ) ) ) AND ( TITLE-ABS-KEY ( reflux OR regurgitation OR gerd OR heartburn ) ) ) AND NOT ( ( TITLE ( case-control OR case-report ) OR TITLE-ABS-KEY ( animal-model* OR mouse OR mice OR murine OR rat OR rats OR rodent* OR porcine OR swine OR pig* OR canine* OR dog* OR baboon* ) AND TITLE ( child* OR pediatric* OR infant* OR newborn* OR neonat* OR girl* OR boy* OR toddler* OR adoles* OR teen* O R youth ) ) ) AND ( LIMIT-TO ( LANGUAGE , “English” ) ) AND ( LIMIT-TO ( DOCTYPE , “ar” ) OR LIMIT-TO ( DOCTYPE , “er” ) ) AND ( LIMIT-TO ( SRCTYPE , “j” ) )

Web of Science Core Collection via Clarivate Analytics (Science Citation Index Expanded 1975+ & Emerging Sources Citation Index 2015+):

#6	#4 NOT #5
#5	((TI=(case-control or case-report )) OR TI=(child* or p$ediatric* or infant* or newborn* or neonat* or girl* or boy* or toddler* or adoles* or teen* or youth)) AND TS=(animal-model* or mouse or mice or murine or rat or rats or rodent* or porcine or swine or pig* or canine* or dog* or baboon*)
#4	#1 AND #2 AND #3 and English (Languages) and Review Article or Meeting Abstract or Book Chapters or Meeting Summary (Exclude - Document Types)
#3	reflux or regurgitation or G?RD or heartburn (Topic)
#2	prevalence* or incidence or morbidit* or comorbid* or epidemiolog* or distribution or population* or cohort* or longitudinal or follow-up or followup or prospective or retrospective or control* or cross-section* or cross-over or random* or trial* or matched-pair* or sampl* or multicenter or study or surveillance or biosurveillance or community-assessment* or common* or frequen* or rate* (Topic) or (single or doubl* or tripl* or treb*) NEAR/3 (blind* or mask*) (Topic)
#1	barrett* or endobrachyoesophagus (Topic) or (columnar or metaplasia or adenocarcinoma or cancer* or carcinoma) NEAR/3 $esophag* (Topic)

Appendix B: Summary of Inclusion and Exclusion Criteria

Inclusion Criteria:

• Sex: Males and Females

• Age: ≥18 years old

• Language: English

• Study includes both patients with and without GERD

• Setting: Any (e.g outpatient setting, endoscopy clinic or inpatient setting)

• Study reports explicitly or implicitly (able to calculate it) the prevalence of BE/EAC in both the GERD and non-GERD groups

• BE is short or long segment diagnosed with EGD or capsule sponge AND pathological confirmation

• GERD was defined/diagnosed by symptoms, validated reflux symptom questionnaires, medication use for GERD, endoscopic assessment or ambulatory pH monitoring

• Study Design: Randomized controlled trials (RCTs), prospective and retrospective cohort studies, case series and cross-sectional studies published in full report in peer-reviewed journals

• Sample size of study (the final sample size; the sample size used for statistics) ≥ 50 patients

Exclusion Criteria:

• Age: <18 years old

• Language: Any language other than English

• Study includes only GERD patients OR only patients without GERD

• GERD and/or BE/EAC not assessed

• Study focuses solely on ultra short segment BE (<1 cm)

• Final sample size < 50 patients

• BE is not diagnosed with EGD or capsule sponge AND pathological confirmation

• GERD was not defined/diagnosed by symptoms, validated reflux symptom questionnaires, medication use for GERD, endoscopic assessment or ambulatory pH monitoring

• Study includes patients with a history of BE undergoing surveillance (without the ability to calculate the proportion of patients diagnosed for the first time)

• Study investigates systemic sclerosis/scleroderma

• Study is a duplicate or uses the same data set as another study

• Full text unavailable

• Study does not report the explicit or implicit prevalence of BE/EAC in both the GERD and nonGERD groups

• Study design: case-control studies, case-reports, reviews (except retrospective chart reviews/retrospective cohort studies), editorials, letters to the editor, conference proceedings/abstracts/theses, book chapters, comments, animal or laboratory studies (e.g., in-vitro or ex-vivo studies), protocols

Appendix C: Modified Version of the Newcastle-Ottawa Scale

The traditional New-Ottawa Scale (NOS) consists of three categories including “Selection”, “Comparability” and “Outcome” which encompass 8 questions and is scored out of 9 points. Specifically, a study can be awarded a maximum of one star for each numbered item within the Selection and Outcome categories. A maximum of two stars can be given for Comparability. We modified the NOS to be more relevant to our specific study, consisting of a total of 7 questions with a maximum score of 8 points. To classify studies as high-quality (low risk-of-bias), medium/acceptable-quality (average risk-of-bias) and low-quality (high risk-of-bias), we calculated the mean and standard deviation (SD) of the modified NOS scores. Taken together, a study score within the range of mean ± 1 SD (4.2 to 6.2, exclusive), ≤ mean - 1 SD (≤ 4.2), and ≥ mean + 1 SD (≥ 6.2), were utilized to categorize studies as of acceptable quality, low quality and high quality, respectively.

Selection

1. Representativeness of the exposed cohort

   1. Multi-center/population-based study 1 (unless c applies)✵
   2. Single-center study (unless c applies) 0.5✵
   3. Selected group of users only (e.g., only obese or smokers or > 50 yo or Caucasian or those with a hiatal hernia or FHx of BE and/or EAC or volunteers or military)
   4. No description of the derivation of the cohort
2. Selection of the non-exposed cohort

   1. Drawn from the same community as the exposed cohort✵
   2. Drawn from a different source
   3. No description of the derivation of the non-exposed cohort
3. Ascertainment of exposure

   1. By symptoms, validated reflux symptom questionnaires, medication use for GERD 1✵
   2. Endoscopic assessment (esophagitis) or ambulatory pH monitoring 0.5✵
   3. No description
4. Demonstration that outcome of interest was not present at start of study

   1. Yes: Index EGD✵
   2. No: Not-index EGD

Comparability

1. Comparability of cohorts on the basis of the design or analysis

   1. Study provides adjusted odds ratio for the risk of BE in GERD patients (e.g., controls for other risk factors of BE) 2✵
   2. Study provides an unadjusted odds ratio for the risk of BE in GERD patients (e.g. controls for other risk factors of BE) 1✵
   3. Study does not provide either a relative risk or odds ratio

Outcome

1. Assessment of outcome: definite histologic confirmation of BE

   1. Confirmed by consensus of 2 expert pathologists or reviewed by 1 expert GI pathologist 1✵
   2. Confirmed by a pathologist but the pathologist type is not stated/community pathologist 0.5✵
2. Outcome Cohort Size

   1. cohort size ≥75 patients with BE 1✵
   2. cohort size between 25 and 74 patients 0.5 ✵
   3. cohort size <25