To the Editor.—The article by Starnes et al1 had three basic flaws, which are as follows.
Barrett's esophagus was not defined. This is important when differentiating cases with tongues of columnar epithelium and exaggerated squamocolumnar junctions (Z lines) from cases with unequivocal Barrett's esophagus. A reasonable approach would be to use the definition, proposed by Skinner et al,2 that at least 3 cm of distal tubular esophagus lined with columnar epithelium be present.
There was a misunderstanding of the terms incidence and prevalence. Prevalence describes the number of cases of a disease that exist at a specific instant in time. Incidence describes the number of new cases of a disease that develop during some specified time interval. To determine the role and frequency of surveillance we need to know the number of new cases of adenocarcinoma that arise in patients who have nondysplastic Barrett's esophagus. The