Endoscopic measurements are frequently used to determine the length of Barrett's epithelium. Decreases in measured length also are used to assess response to pharmacotherapy. The aim of this study was to determine the accuracy of the endoscopic measurement of length under standardized conditions.

A life-size model of the esophagus was built by using an opaque polyvinyl tube. Transparent plastic inserts designed to resemble Barrett's epithelium were created. Ten different insert lengths were randomly studied. Trained endoscopists and fellows in training blinded to the lengths of the inserts performed measurements twice in random order.

A total of 240 measurements were made by 12 endoscopists. The mean difference (1 standard deviation) between the measured and correct length was 1.10 (1.7) cm. Overestimation of the length occurred in 47% and underestimation in 37%. The kappa statistic for the first and second measurement (within 0.5 cm) performed by the same individual (intra-observer) was 0.40, suggesting only fair agreement. There was considerable variability between endoscopists.

Accurate measurement of length at endoscopy is difficult even under ideal conditions. Intra-observer agreement is fair, but results obtained by different endoscopists are widely divergent. Small improvements in the length of Barrett's epithelium observed in some clinical trials could be because of chance instead of therapeutic effect.