Ambulatory Esophageal Manometry: Comparison of Expert and Computer‐Aided Analyses

This study describes the validation of a computer program for automated analysis of ambulatory 24‐hour two‐channel esophageal manometry. The program's ability to characterize contractions and to calculate their duration and amplitude was validated against manual evaluation. An independent reference standard for the identification and classification of contractions was established by submitting representative recordings to a group of 14 experts in upper gastrointestinal motility; the program's ability to identify and classify contractions was then validated against the majority verdict of the experts. The results show an excellent correlation between the manual and computer evaluations of both contraction amplitude (p = 0.9957) and duration (p = 0.8241). The concordance between the experts was also excellent: 97 (72%) of 135 pressure events were classified identically by 12 or more experts. Computer‐aided manometry analysis (CAMA) had a sensitivity of 98.9% and a specificity of 93.5% for the detection of contractions, with sensitivities of 94.7% and 84.4% and specificities of 94.9% and 94.2% for the classification of propagated and nonpropagated contractions, respectively. Copyright © 1990, Wiley Blackwell. All rights reserved