Risk factors for the detection of Barrett's esophagus in patients with erosive esophagitis

BackgroundBarrett's esophagus (BE) results from metaplastic healing of injured esophageal mucosa after erosive esophagitis (EE).ObjectiveOur purpose was to determine whether severity of esophagitis, indication for endoscopy, or proton pump inhibitor treatment affects the subsequent incidence of BE diagnosis in patients found to have EE on EGD performed for any indication.DesignWe identified patients with primary or secondary International Classification of Diseases, 9th revision diagnosis codes of EE from 1996 to 2006 who had at least 2 EGDs on record. Patients with prevalent BE on the first EGD were excluded.SettingInpatients and outpatients at Stanford University and Palo Alto Veterans Affairs Health Care System.InterventionsRetrospective review of EGD and pathology reports to confirm BE.Main Outcome MeasurementsDetection of BE after diagnosis of EE.ResultsA total of 1095 patients were identified between 1996 and 2000, and 102 (9%) were included. Sixty-two (61%) patients were veterans, 87 (85%) were male, and 83 (81%) were white. The mean (±SD) age was 58 ± 14 years (range 24-83 years). BE was detected in 9 (9%) patients (95% CI, 4.5%-17.6%) over a mean of 13.3 ± 5.7 months (range 1-53.5 months), and all had prior grade 4 esophagitis. The mean BE length was 4 ± 1.8 cm (range 1-18 cm). Six patients had upper GI bleeding as the indication for EGD, whereas the other 3 complained of dysphagia. The association of grade 4 esophagitis (P = .01) and GI hemorrhage (P = .01) to the subsequent detection of BE was highly statistically significant.LimitationsRetrospective study, small number of patients with BE after EE. All patients were receiving care at tertiary medical centers.ConclusionsBE was detected in 9% of patients with prior EE and was detected exclusively on follow-up of patients with severe esophagitis. The majority of the patients found to have BE had upper GI bleeding as the presentation for EGD.