Adherence to Surveillance Guidelines in Nondysplastic Barrett's Esophagus

Introduction: Surveillance patterns in Barrett's esophagus (BE) are not well characterized. Guidelines published between 2002 and 2008 recommended surveillance esophagogastroduodenoscopy (sEGD) at 3-year intervals for nondysplastic BE (NDBE). We assessed guideline adherence in incident NDBE in a Veterans Affairs (VA)-based study. Methods: At a single VA center, we identified incident cases of biopsy-confirmed NDBE between January, 2006 and December, 2008. We excluded patients aged 76 years and above and those who developed BE-associated dysplasia or cancer during follow-up. All sEGDs through October, 2014 were documented. Our primary criteria classified cases as guideline adherent if a sEGD was performed within 6 months of each expected 3-year surveillance interval; in cases with 2 sEGDs, 1 sEGD >6 months, and 1 year outside an interval was allowed if the average interval was between 2.5 and 3.5 years. Comorbidity, primary care encounters, presence of long-segment BE (LSBE), endoscopist recommendations, and Charlson comorbidity index (CCI) were assessed. Results: We identified 110 patients (96.4% male, 93.6% white) with mean age 58.98.5 years at index EGD. Median follow-up was 6.7 years (range, 3.7 to 8.6). Thirty-three (30.0%) cases were guideline adherent; 77 (70.0%) cases were nonadherent, including 52 (47.3%) with irregular surveillance and 25 (22.7%) with no surveillance. Forty cases (14 adherent) had 1 sEGD, 36 (18 adherent) had 2, 8 (1 adherent) had 3, and 1 nonadherent case had 4. Adherent cases were significantly older (61.5 vs. 57.9 y, P = 0.04), and tended to have more LSBE (33.3% vs. 20.8%, P = 0.16). There were no differences between adherent and nonadherent cases in annual primary care encounters (72.7% vs. 66.2%, P = 0.66), CCI >= 4 (15.2% vs. 15.6%, P = 0.95), biopsy-positive sEGDs (75.8% vs. 76.6%, P = 0.92), and any recommendation for subsequent surveillance (81.8% vs. 77.9%, P = 0.65). A logistic regression model using age, CCI, and LSBE showed an independent association between adherence and older age (P = 0.03). Conclusions: In a single-center VA cohort, sEGD of NDBE was mostly nonadherent to guidelines. Adherent cases were older at baseline with a trend toward more LSBE. A larger study is needed to identify medical and social factors associated with adherence or nonadherence to surveillance.