Prospective blinded trial of a fecal antigen test for the detection of Helicobacter pylori infection.

Abstract

We prospectively studied an FDA-approved stool test for the diagnosis of Helicobacter pylori infection based on the detection of antigens shed in stool. We studied 108 patients undergoing endoscopy. Stool samples were collected on the day of the endoscopy and read in a blinded manner by a technician who was unaware of the findings of the endoscopy. The results of stool testing were compared to those of the rapid urease test (Pyloritek). There were 51 patients infected with H. pylori by rapid urease testing and 57 who were not. The sensitivity of the stool test was 86% (95% confidence interval: 77-96%), the specificity was 91% (81-97%), with a positive predictive value of 89% (77-96%) and a negative predictive value of 88% (81-97%). The likelihood ratio for a positive test was 10.18 and the likelihood ratio for a negative test was 0.11. The stool test was sensitive and specific for the detection of H. pylori in a U.S. population.