LOW PREVALENCE OF ENDOSCOPIC SCREENING FOR BARRETT’S ESOPHAGUS IN A SCREEN-ELIGIBLE PRIMARY CARE POPULATION

Abstract

Objectives:

Despite societal recommendations supporting Barrett’s esophagus (BE) screening, it is unknown what proportion of eligible patients are screened in primary care. We assessed the proportion of BE screen-eligible patients evaluated in the primary care setting receiving upper esophagogastroduodenoscopy (EGD), and to identify factors associated with undergoing EGD.

Methods:

This was a retrospective study of BE screen-eligible patients, as defined by the American College of Gastroenterology’s BE guidelines, in a multi-practice healthcare network consisting of 64 internal medicine (IM) and 94 family medicine (FM) practices. The proportion undergoing EGD, prevalence of BE and esophageal adenocarcinoma (EAC) in this group, and patient and provider factors associated with undergoing EGD were assessed. Multivariable logistic regression was performed to identify independent predictors of undergoing EGD.

Results:

Of 1127 screen-eligible patients, mean age was 65.2±8.6 years, 45% were obese, and 61% were smokers. 73% were seeing FM, 94% were on proton pump inhibitors and 44% took ≥1 GERD medication. Only 39% (n=436) of patients had undergone EGD. Overall prevalence of BE or EAC was 9.9%. Of 39 (9%) referred for BE screening as the primary indication, BE/EAC prevalence was 35.1%. Factors associated with increased odds of having EGD were symptomatic GERD despite treatment (OR [95% CI]=12.1[9.1–16.3]), being on ≥1 GERD medication (OR [95% CI]=1.4[1.0–1.9]), and being an FM patient (OR [95% CI]=1.5[1.1–2.1]).

Conclusions:

In this large, primary care population, only 39% of screen-eligible patients underwent EGD. A majority of exams were triggered by refractory symptoms rather than screening referrals, highlighting a need for improved dissemination and implementation of BE screening.

INTRODUCTION

Esophageal adenocarcinoma (EAC) has the most rapidly rising incidence of any cancer in the United States (US), with a six-fold increase in the last 40 years (1). Outcomes remain dismal, with a median survival of <1 year following diagnosis. Survival rates are better if EAC is diagnosed at an earlier, asymptomatic stage (2, 3). One approach to early EAC detection is to identify Barrett’s esophagus (BE), a well-recognized precursor to EAC. Because 60% of EAC cases are diagnosed in GERD patients (4), targeting screening to high-risk individuals with GERD has the highest yield in terms of earlier EAC detection. While BE screening is not recommended by the United States Preventive Services Taskforce (USPSTF) due to lack of randomized controlled trial data, it is widely endorsed by multiple guidelines in the United States and around the world (5–13), including the American College of Physicians’ (ACP) Best Practice Advice (11) and the American College of Gastroenterology (ACG) BE guidelines (12, 13). The newest ACG guidelines (12) recommend BE screening with esophagogastroduodenoscopy (EGD) in patients with chronic or symptomatic GERD and three or more additional risk factors such as male sex, age ≥50 years, Caucasian race, smoking, obesity, and first-degree relative with BE or EAC.

Despite these screening recommendations, >85% of EAC patients have never had an EGD prior to their cancer diagnosis (14, 15), and only approximately 8% of the >10,000 cases of EAC diagnosed annually in the US are detected through current approaches of cancer prevention (4). This indicates that high-risk patients are ineffectively screened. Prior work has shown that gastroenterologists (GI) are highly likely to perform EGD in symptomatic or chronic GERD patients (16), with one study (17) showing that referral for EGD occurred in 51% of symptomatic GERD care episodes by GI compared to 6% of GERD care episodes by primary care providers (PCPs). The large discrepancy between referrals for EGD between PCPs and GIs suggest that optimizing screening at the patient or primary care level is an unrealized opportunity to improve EAC detection, as a vast majority of GERD patients have their health care interactions in primary care. However, data documenting the proportion of screen-eligible patients who undergo endoscopic screening for BE are lacking. Such data are crucial to understand the impact of interventions designed to increase endoscopic screening in appropriate populations.

Therefore, the aims of this study were to: 1) to assess the proportion of patients eligible for BE screening who have undergone an EGD in a large primary care network, and 2) to identify patient- and provider-associated factors associated with the use of EGD.

MATERIALS AND METHODS

Study Design and Data Source

This is a retrospective study of adult patients conducted in a large healthcare network consisting of 64 internal medicine (IM) and 92 family medicine (IM) practices in both academic- and community-based settings throughout North Carolina. Electronic health record (EHR) data were obtained from the Carolina Data Warehouse for Health (CDW-H), a central data repository within the University of North Carolina (UNC) Healthcare system developed for research purposes. Data from UNC EHRs are transferred to the CDW-H and updated on a nightly basis. In addition, clinical data from any scanned medical records and data from all institutions in the EPIC Care Everywhere network, are also available. In contrast to insurance-based administrative data, CDW-H data are highly granular, and all detailed clinical information can be accessed. The study was approved by the University of North Carolina institutional review board.

Study Population

For this study, we focused on BE screening-eligible patients according to the American College of Gastroenterology’s BE guidelines (13), published in 2016. These guidelines recommend screening in men with chronic and/or symptomatic GERD and two or more risk factors for BE or EAC which include age >50 years, Caucasian race, presence of central obesity, current or past history of smoking, and a confirmed family history of BE or EAC (in a first-degree relative).

Based on these guidelines, we defined the index screening population as any adult, Caucasian, male >50 years of age who obtained continued outpatient medical services in the UNC primary care network from 2016 to 2018 for a primary or secondary diagnosis of GERD. No other risk factors were used to define the screen-eligible group. We chose age, male sex and Caucasian race as the three additional BE/EAC risk factors as these are widely recognized, strong risk factors, (18) which are accurately documented in EHR data compared to other risk factors such as family history and centripetal obesity.

Case Identification

The case definition for GERD was based on prior studies (17, 19) and defined as: (1) ≥2 consecutive primary or secondary diagnoses for GERD; and (2) ≥1 medication prescription for either a proton pump inhibitor (PPI), H2RA, other anti-reflux therapy including OTC agents. The date of the first relevant medication prescription was defined as the index date; (3) Diagnosis of GERD during the 60 days before and after the index date. Subjects were also required to have continuous enrollment in the system for at least 6 months before and at least 1 year after the index date. Patients >80 years of age, achalasia, eosinophilic esophagitis (EoE), Zollinger-Ellison syndrome, anti-reflux surgery, total or partial esophagectomy were excluded, as were patients with a prior history of BE undergoing surveillance.

After initial cohort identification though the CDW-H data query, detailed manual chart review was additionally performed of all patients to confirm screen eligibility status and other inclusion criteria. The performance of an EGD was determined by chart abstraction, searching for EGD in any EHR data, including any scanned records from other institutions, EGD reports from community gastroenterologists not within the health system, and nationwide institutions in the EPIC Care Everywhere network. Detailed chart abstraction was performed to collect variables of interest. The diagnosis of BE was confirmed by endoscopic reports demonstrating at least 1 cm of columnar epithelium in the tubular esophagus, and corresponding pathology from that mucosa confirming intestinal metaplasia (IM).

Statistical Analysis

Bivariate analyses were performed to determine baseline differences in demographics, symptoms, clinical features, and primary provider type using Student’s t-test for continuous variables and Pearson’s chi-square tests for categorical variables. If data were not normally distributed, non-parametric techniques were used. Multivariable logistic regression was performed to identify independent predictors of those who underwent EGD. Risk factors for undergoing EGD on bivariate comparison with p <0.10 were included in the initial model, taking into account collinear and confounding variables. The initial model was subsequently reduced until all factors were significant at the p <0.05 level. The model was further simplified to include only the variables that were significantly contributing to the model using likelihood ratio testing. We also calculated the yield of BE detection in subgroups with additive risk factors such as smoking and obesity compared to the baseline population of Caucasian men >50 years of age. All analyses were performed using Stata 13.1 (StataCorp LP, College Station, TX). Statistical significance was determined at α = 0.05, and all p values for statistical significance were 2-sided.

RESULTS

There were 1154 subjects identified by the final search query of which 27 were excluded after manual chart review (Figure 1), yielding 1127 subjects in the final sample. Of 1127 subjects, mean age was 65.2 ± 8.6 years, 45% has BMI >30, and 61% were former or current smokers (Table 1). A majority were married, unemployed/retired, and had Medicare as their primary insurance coverage. Seventy-three percent were seeking primary health services under FM. Based on the inclusion criteria, all were on some type of prescription or OTC GERD medication with 94% on PPI therapy and 44% maintained on more than one medication for GERD. A majority had no to minimal (less than weekly) heartburn and reflux symptoms on medical therapy, with 43% reporting breakthrough heartburn and reflux symptoms on medical treatment.

Figure 1:

Flow Diagram of Proportion undergoing Upper Endoscopy and Prevalence and Characteristics of Barrett’s Esophagus

Table 1:

Population Characteristics

	Total Sample (n= 1127)
Age, mean ± SD	65.2 ± 8.6
51–59	335 (29.7)
60–69	383 (34.0)
70–80	410 (36.3)
Ethnicity, n (%)
Hispanic or Latino	13 (1.2)
Not Hispanic or Latino	1085 (98.8)
BMI, mean ± SD	30.1 ± 5.9
<24.9	192 (17.0)
25.0–29.9	431 (38.3)
>30	504 (44.7)
Smoking Status, n (%)
Current	181 (16)
Former	507 (45)
Never	439 (39)
Employment status, n (%)
Employed	417 (42.8)
Unemployed/Retired	555 (57.2)
Marital status, n (%)
Married	771 (71.9)
Not married	301 (28.1)
Insurance Status, n (%)
No insurance	53 (4.7)
Private	443 (39.3)
Medicare	599 (53.2)
Medicaid	32 (2.8)
Primary Care Specialty, n (%)
Internal Medicine	302 (26.8)
Family Medicine	825 (73.2)
Procedures, n (%)
EGD	436 (38.7)
pH impedance testing	11 (0.98)
Esophageal manometry	16 (1.4)
GERD medications, n (%)
PPI	1056 (93.7)
H2-blockers	408 (36.2)
Baclofen	26 (2.3)
Carafate	128 (11.4)
Over-the-counter agents	293 (26.0)
Use of >1 medications	495 (43.9)
GERD Symptoms on Treatment, n (%)	488 (43.3)

Of all the patients eligible for BE screening, only 38.7% (n=436) underwent an EGD for any indication, while 61.3% (n=691) never had an EGD (Figure 1). Among those who underwent an EGD, the most common procedure indications were symptomatic GERD (54%), dysphagia (25%), abdominal pain (16%), bleeding (16%), and other reasons (17%) (Figure 2). Among all patients who underwent EGD, the prevalence of BE (n=38) or EAC (n=3) was 9.9% (41/436). On EGD, 23.8% had esophagitis, 20.4% had an irregular Z-line with no associated IM, and 26.7% had a hiatal hernia (Figure 2). Only 9.2% (n=39) were referred for Barrett’s screening as a primary indication for the EGD; in this group, the prevalence of BE or EAC was 35.1%.

Figure 2:

Indication* for Upper Endoscopy and Endoscopic Findings

Among the 41 patients with BE or EAC, 24.4% (n=10) had long segment BE, 75.6% (n=31) had short segment BE, and 7.3% (n=3) had EAC (Table 2). None of the BE cases had low-grade or high-grade dysplasia. The most common indications for EGD in those with BE or EAC were symptomatic GERD (46%), Barrett’s screening (32%), dysphagia (22%), and gastrointestinal bleeding or anemia (20%).

Table 2:

Characteristics of those with Barrett’s esophagus or esophageal adenocarcinoma on endoscopy

	Sample (n=41)
Age, mean ± SD	64.4 ± 7.8
BMI, mean ± SD	29.7 ± 5.0
History of smoking, n (%)	30 (73.2)
Use of >1 acid suppressive medications, n (%)	19 (46.3)
EGD indication, n (%)
GERD	19 (46.3)
Chest pain	0 (0)
Dysphagia	9 (21.9)
Cough	3 (7.3)
Bleeding/Anemia	8 (19.5)
Barrett’s Screening	13 (31.7)
Abdominal Pain	7 (17.1)
Other	7 (17.1)
EGD findings, n (%)
Long Segment BE	10 (24.4)
Short Segment BE	31 (75.6)
Hiatal Hernia	14 (34.2)
Esophagitis	8 (19.5)
Esophageal Stricture	5 (12.2)
Pathology, n (%)
Non-dysplastic BE	38 (92.7)
Low-grade dysplasia	0 (0)
High-grade dysplasia	0 (0)
Intramucosal adenocarcinoma	1 (2.4)
≥ Intramucosal adenocarcinoma	2 (4.9)

Features significantly associated with having an EGD were a lower mean BMI (29.6 ± 5.8 vs. 30.4 ± 5.9), being married (74% vs. 68%), and being under the care of a FM versus IM provider (67% vs, 34%); p<0.05 for all (Table 3). A higher proportion of those who underwent EGD were also on GERD medications that included both prescription and OTC drugs (52% vs. 39%), and were also more likely to be symptomatic with breakthrough GERD symptoms despite maintenance therapy (78% vs. 22%), p<0.01 for both. In multivariable logistic regression, patient features associated with an increased odds of undergoing an EGD were having symptomatic GERD despite medical treatment (OR [95% CI] = 12.1 [9.1–16.3]) and being on more than one GERD medication concurrently (OR [95% CI] = 1.4 [1.0–1.9]). In terms of provider features, being under the care of a FM provider resulted in an increased odds (OR [95% CI] = 1.5 [1.1–2.1]) of undergoing an EGD compared to an IM provider.

Table 3:

Comparison of clinical characteristics of patients with and without an EGD

	No EGD (n=694)	EGD (n=433)	p value
Age, mean ± SD	65.5 ± 8.5	64.7 ± 8.7	0.13
51–59	201 (29.0)	134 (31)	0.71
60–69	234 (33.7)	149 (34.4)
70–79	246 (35.4)	140 (32.3)
80+	13 (1.9)	10 (2.3)
Hispanic or Latino, n (%)	5 (0.7)	8 (1.9)	0.09
BMI, mean ± SD	30.4 ± 5.9	29.6 ± 5.8	0.04
<24.9	112 (16.1)	80 (18.5)
25.0–29.9	259 (37.3)	172 (39.7)
>30	323 (46.6)	181 (41.8)
Smoking Status, n (%)			0.48
Current	109 (15.7)	72 (16.6)
Former	322 (46.4)	185 (42.7)
Never	263 (37.9)	176 (40.7)
Employment status, n (%)			0.59
Employed	259 (43.5)	156 (41.7)
Unemployed	337 (56.5)	218 (58.3)
Marital status, n (%)			0.02
Married	488 (74.4)	283 (68.0)
Not married	168 (25.6)	133 (33.0)
Insurance Status, n (%)			0.58
No insurance	34 (4.9)	19 (4.4)
Private	274 (39.5)	169 (39.0)
Medicare	370 (53.3)	229 (52.9)
Medicaid	16 (2.3)	16 (3.7)
Primary Care Specialty, n (%)			<0.01
Internal Medicine	157 (22.6)	145 (33.5)
Family Medicine	537 (77.4)	288 (66.5)
GERD medications, n (%)
PPI	640 (92.2)	416 (96.1)	0.01
H2-blockers	219 (31.6)	189 (43.6)	<0.01
Baclofen	16 (2.3)	10 (2.3)	0.99
Carafate	40 (4.8)	88 (20.3)	<0.01
Over-the-counter agents	157 (22.6)	136 (31.4)	<0.01
Use of >1 medications	268 (38.6)	227 (52.4)	<0.01
GERD Symptoms on Treatment, n (%)	151 (21.8)	337 (78)	<0.01

The yield of finding BE or EAC performed in the base study population of Caucasian men, >50 years of age with GERD was 9.9 cases/100 upper endoscopies (Table 4). When additive BE risk factors (smoking and BMI) to the baseline study population (Caucasian men, >50 years of age with GERD) were assessed, the yield of detecting BE/EAC per 100 upper endoscopies in overweight smokers was three times higher than in the non-smoker group with a normal BMI.

Table 4:

Yield of finding Barrett’s esophagus per 100 upper endoscopies based on additive body mass index (BMI) and smoking status risk factors to the baseline study population of Caucasian men, >50 years of age with gastroesophageal reflux disease (GERD).

	Body mass index (BMI) Categories
	Normal	Overweight	Obese	Overweight/Obese	All
Never Smoker	7.69 [0.95, 25.13]	7.89 [2.95, 16.40]	4.48 [0.93, 12.53]	6.29 [2.92, 11.61]	6.51 [3.29, 11.35]
Former Smoker	6.90 [0.85, 22.77]	10.14 [4.18, 19.79]	13.58 [6.98, 23.00]	12.00 [7.27, 18.30]	11.17 [6.96, 16.73]
Current Smoker	8.33 [1.03, 27.00]	22.73 [7.82, 45.37]	13.64 [2.91, 34.91]	18.18 [8.19, 32.71]	14.71 [7.28, 25.39]
Ever Smoker	7.55 [2.09, 18.21]	13.19 [7.00, 21.90]	13.59 [7.63, 21.75]	13.40 [8.95, 19.02]	12.15 [8.35, 16.88]
All	7.59 [2.84, 15.80]	10.78 [6.51, 16.50]	10.00 [5.93, 15.53]	10.39 [7.34, 14.15]	9.86 [7.17, 13.13]

DISCUSSION

While multiple societal guidelines (5–11, 13), as well as ACP Best Practice Advice (11), have supported endoscopic screening for BE in high-risk patients, it is unknown how often this guidance is applied in the primary care setting. In this study of BE screen-eligible patients in a large primary care network, only 39% had ever had an EGD for any reason. Those who had an EGD were more likely to be seen by a FM than an IM provider, be on multiple GERD medications, and have symptomatic GERD on medical therapy. In those who had an EGD, the yield for BE/EAC was substantial: 9.1% had BE and 0.7% had EAC. These data suggest that screening EGD is underutilized in populations who may be appropriate for this exam. It also suggests that inadequate control of reflux symptoms, as opposed to an active decision to screen the patient, is the primary motivator for these exams. Of note, the definition of chronic GERD symptoms used in this study yielded patients with relatively severe disease, as evidenced by the proportion of symptomatic patients despite therapy and the number of patients on more than one anti-secretory agents. Therefore, the 39% screening rate observed in our study, while disappointing, may actually represent an overestimate of screening in a more general GERD population, if one presumes that patients with less severe GERD symptoms would be less likely to be screened. Since we chose obvious, well-recognized risk factors for BE in order to fulfill guideline criteria for screening, it is reasonable to believe that the findings in this study may actually represent a “best-case” scenario.

This study addresses an important question: What proportion of screening eligible patients are actually screened for BE in primary care. Our data are unique in that they represent a population-based cohort of screen-eligible patients in a non-veteran primary care setting. A recent retrospective study (20) of 182 veterans with EAC found that 36.5% of a high-risk GERD group had an EGD prior to their EAC diagnosis, which is similar to the proportion in our non-veteran sample. Previous studies (14, 15) have focused on subjects who had already developed EAC, asking what proportion of cancer sufferers had a screening exam prior to developing cancer. These studies uniformly suggest that screening EGD was rare in this population. Consequently, in one systematic review (21), only 5% of patients with EAC had a known diagnosis of BE. While these studies are illuminating, they focus only on those who have already suffered the most dreaded complication of GERD. They also include patients who do not meet criteria for BE screening due to lack of BE/EAC risk factors. Our study instead focuses on the “top of the funnel”: the more difficult-to-capture population of patients who are screen-eligible. We ask the question of whether there is room for improvement in identifying these high-risk patients in primary care to help avert EAC.

There are multiple potential reasons why such a small proportion of EAC patients undergo EGD and have known BE. One of the reasons is that there is a proportion of BE and EAC patients who do not have typical GERD symptoms (22), and thus never seek clinical care for this condition until becoming symptomatic. Such patients would never be diagnosable with BE prior to their cancer diagnoses by using current guidelines. A second reason is that some EACs may arise from minute segments of BE, and thus, not be detected at the time of a screening upper endoscopy. However, an important, more actionable, reason is that high-risk screen-eligible patients may not be recognized or referred for BE screening in the primary care setting, where a majority of longitudinal care occurs. Our findings show that about two-thirds of patients seen in primary care with GERD and additional risk factors for BE/EAC never undergo screening, highlighting a significant missed opportunity for BE screening in the primary care setting. The results of this study conflict with a prior report (23) suggesting that internists were more likely to refer for EGD for GERD symptoms when compared to FM providers. Since internists in our system tend to take care of patients with higher chronic morbidity, the decreased screening rates seen in the IM cohort may represent more selective screening in a less healthy population. Finally, although screening EGD for patients such as those in our cohort is recommended by multiple professional societies, there is overall limited guidance on this topic in the general medicine literature, other than the ACP Best Practice Advice from 2012. Endoscopic screening of GERD patients has not been endorsed by the US Preventive Services Task Force and, it may be that the lack of such an endorsement is impactful on BE screening patterns in primary care. Although our data cannot address this question, it highlights a need for higher quality BE screening studies in the future and updated practice advice for the primary care setting.

One of the risk factors for failing to undergo a screening EGD was having well-controlled GERD symptoms on maintenance PPI therapy, which renders this group as “asymptomatic,” or nearly so. Complications of chronic GERD such as BE and EAC are likely not discussed in clinical encounters in those who have well-controlled GERD and are asymptomatic on therapy, and thus opportunities for screening referrals are missed. Further studies are currently underway to delineate primary provider perspectives and barriers for BE screening. Unlike screening guidelines for colorectal cancer, which are based on age, there are a number of clinical factors that are incorporated into risk stratifying GERD patients (13). Similar to existing literature (18, 24, 25), our study results demonstrate an additive yield of BE detection in smokers who are overweight or obese. Risk stratification calculators are now available which take into account the incremental yield of BE detection with additive risk factors (26, 27), and should be utilized to perform targeted screening to identify those at highest risk. Furthermore, incorporation of these risk stratification calculators into EHR systems or creation and implementation of Best Practice Advisory computer prompts may be beneficial to serve as triggers for screening during clinical encounters. Risk stratification may also aid in guiding the choice of screening modality with lower risk individuals potentially screened with non-endoscopic screening tools, which have recently been reported to have acceptable performance characteristics and to increase the yield of new diagnoses of BE, when compared to standard of care endoscopic screening (28, 29).

This study has multiple strengths. It is one of the first studies to assess the prevalence of BE screening in a non-veteran primary care population, which is presumably more reflective of the general BE population. The sample size is large, and included patients seen in community and academic practices by FM and IM practitioners. Detailed manual chart abstraction was performed of all cases identified through the EHR, which included review of all scanned documents and records from institutions that were not part of the healthcare network and thorough clinical note review. The study also delineates the incremental yield of EGD for BE detection with additive risk factors, on which there are limited data available. Some limitations are that, although it was a large primary care population from a geographically dispersed population, it may not be a nationally representative sample of patients, and screening rates may differ by region or healthcare system. We chose widely recognized BE risk factors, meaning that our selected population was “screenable,” but certainly did not include the entire universe of GERD patients in our patient population, since we wanted to assure the group we studied fulfilled screening criteria. As previously mentioned, the definition of GERD used yielded patients with severe disease, leading to a potential overestimate of screening rates. Despite the detailed and comprehensive medical record review, some EGDs performed out of network may not have been captured. Finally, due to the retrospective nature of the study, and our use of EHR data to define our initial potential screenable population, we were not able to identify specific systematic and individual level reasons for not undergoing BE screening, including non-adherence to a recommended screening referral. Future qualitative studies addressing this specific question at both a patient and physician level would be valuable and are underway.

In summary, of BE screen-eligible patients seen in a primary care setting, only 39% underwent an EGD for any reason. The yield of BE or EAC in this population was 10%. Patients who had an EGD were referred to primarily investigate symptoms such as medically refractory GERD or dysphagia. Only 10% of those who had an EGD were referred for BE screening, in whom the prevalence of BE or EAC was 35%. This study highlights opportunities for improvement in identifying and screening those patients at highest risk for developing BE and EAC in primary care. The yield of BE detection increased with the additive risk factors of smoking and increased BMI. Therefore, targeting patients for screening with multiple known risk factors may result in the highest reward in terms of earlier EAC detection and prevention.

STUDY HIGHLIGHTS.

WHAT IS KNOWN

• One approach to early detection of esophageal adenocarcinoma (EAC) is to identify it’s precursor, Barrett’s esophagus (BE).

• Because a majority of EAC cases are diagnosed in patients with gastroesophageal reflux disease (GERD), guidelines recommend targeting BE screening in high-risk individuals with GERD and additional risk factors.

• Despite these screening recommendations, >85% of EAC patients have never had an EGD prior to their cancer diagnosis and data documenting the proportion of screen-eligible patients in primary care who undergo endoscopic screening for BE are lacking.

WHAT IS NEW HERE

• This one of the first studies to assess the prevalence of BE screening in a non-veteran primary care population, which is likely more reflective of the general BE population.

• Only 39% of BE screen-eligible patients underwent an EGD for any reason, of whom 10% had BE or EAC. Prevalence of BE/EAC was 35% in those who had an EGD primarily for BE screening.

• This study highlights opportunities for improvement in identifying and screening those at highest risk for developing BE and EAC in primary care.