Practice patterns of surveillance endoscopy in a Veterans Affairs database of 29,504 patients with Barrett's esophagus.

BACKGROUND

Practice guidelines recommend surveillance endoscopy every 2 to 3 years among patients with Barrett's esophagus (BE) to detect early neoplastic lesions. Although surveys report that >95% of gastroenterologists recommend or practice BE surveillance, the extent and patterns of surveillance in clinical practice are unknown.

OBJECTIVE

To identify the extent and determinants of endoscopic surveillance among BE patients.

DESIGN

Retrospective cohort study.

SETTING

A total of 121 Veterans Affairs facilities nationwide.

PATIENTS

Veteran patients with BE diagnosed from 2003 to 2009, with follow-up through September 30, 2010.

INTERVENTION

Not an interventional study.

MAIN OUTCOME MEASUREMENTS

The proportions of patients with BE who received any EGD after the index BE EGD date. In the subgroup of patients with at least 6 years of follow-up, we also calculated proportions for regular (EGD during both 3-year intervals), irregular (EGD in only 1 interval), and no surveillance. We examined differences in demographics and clinical and facility factors among these groups in unadjusted and adjusted analyses.

RESULTS

We identified 29,504 patients with BE; 97% were men, 83% white, and their mean age was 61.8 years. During a 3.8-year median follow-up period, 45.4% of patients with BE received at least one EGD. Among the subgroup of 4499 patients with BE who had at least 6 years of follow-up, 23.0% had regular surveillance, and 26.7% had irregular surveillance. There was considerable facility-level variation in percentages with surveillance EGD across the 112 facilities and by geographic region of these facilities. Demographic and clinical factors did not explain these variations. Patients with at least one EGD were significantly more likely to be white; to be aged <65 years, with a low level of comorbidity; to have GERD, obesity, dysphagia, or esophageal strictures; to have more outpatient visits; and to be seen in smaller hospitals (<87 beds) than those without any EGD.

LIMITATIONS

There might be misclassification of BE and surveillance EGD. Lack of pathology data on dysplasia, which dictates surveillance intervals.

CONCLUSION

Endoscopic surveillance for BE is considerably less commonly practiced in Veterans Affairs facilities than is self-reported by physicians. Although several clinical factors are associated with variations in surveillance, facility-level factors play a large role. The comparative effectiveness of the different practice-based surveillance patterns needs to be examined.