Randomized crossover study comparing efficacy of transnasal endoscopy with that of standard endoscopy to detect Barrett's esophagus

Background: Unsedated transnasal endoscopy (TNE) may be safer and less expensive than standard endoscopy (SE) for detecting Barrett's esophagus (BE). Emerging technologies require robust evaluation before routine use.

Objective: To evaluate the sensitivity, specificity, and acceptability of TNE in diagnosing BE compared with those of SE.

Design: Prospective, randomized, crossover study.

Setting: Single, tertiary-care referral center.

Patients: This study enrolled consecutive patients with BE or those referred for diagnostic assessment.

Intervention: All patients underwent TNE followed by SE or the reverse. Spielberger State-Trait Anxiety Inventory short-form questionnaires, a visual analogue scale, and a single question addressing preference for endoscopy type were administered.

Main outcome measurements: Diagnostic accuracy and tolerability of TNE were compared with those of SE.

Results: Of 95 patients randomized, 82 completed the study. We correctly diagnosed 48 of 49 BE cases by TNE for endoscopic findings of columnar lined esophagus compared with the criterion standard, SE, giving a sensitivity and specificity of 0.98 and 1.00, respectively. The BE median length was 3 cm (interquartile range [IQR] 1-5 cm) with SE and 3 cm (IQR 2-4 cm) with TNE, giving high correlations between the two modalities (R(2) = 0.97; P < .001). The sensitivity and specificity for detecting intestinal metaplasia by TNE compared with those by SE was 0.91 and 1.00, respectively. The mean (± standard deviation) post-endoscopy Spielberger State-Trait Anxiety Inventory short-form score for TNE (30.0 ± 1.10 standard error of the mean [SEM]) was lower than that for SE (30.7 ± 1.29 SEM), (P = .054). The visual analogue scale scores were no different (P = .07). The majority of patients (59%) expressed a preference for TNE.

Limitations: This is a small study, with limited generalizability, a high prevalence of patients with BE, differential drop-out between the two procedures, and use of sedation.

Conclusion: TNE is an accurate and well-tolerated method for diagnosing BE compared with SE. TNE warrants further evaluation as a screening tool for BE.