Diagnostic tests for Helicobacter pylori: a prospective evaluation of their accuracy, without selecting a single test as the gold standard.

To assess the accuracy of six commonly used diagnostic tests for Helicobacter pylori in a prospective study without using any specific test as the gold standard (the patient was regarded as H. pylori-infected if two or more tests, whatever their nature, were positive).In 105 outpatients undergoing upper GI endoscopy, 62 without significant abnormalities, 28 with gastroesophageal reflux disease, 19 with peptic ulcer, one with erosive gastritis, and one with atrophic gastritis (some patients had more than one diagnosis), antral biopsy specimens were taken for culture, polymerase chain reaction, histological examination (hematoxylineosin and Giemsa stains), and rapid urease test. Corpus biopsy specimens were taken for histological examination. Serology (ELISA) and a 13C-urea breath test were also performed. Consistency of diagnosis between two pathologists was assessed by kappa statistics.Sensitivity and specificity, respectively, were as follows: culture, 98.4 and 100%; polymerase chain reaction, 96.7 and 100%; histological examination (antrum), 96 and 98.8%; histological examination (antrum + corpus), 98.4 and 98.8%; rapid urease test, 90.2 and 100%; 13C-urea breath test, 100 and 100%; and serological examination, 98.4 and 88.4% (95% in those who had not been previously treated for H. pylori). All H. pylori-positive cases were detected by culture and rapid urease test. In 86.4% of these cases all antral biopsy-based tests were positive. Agreement between pathologists was good, with a kappa coefficient around 0.90 for antral biopsy specimens.All antral biopsy-based tests, as well as the 13C-urea breath test, are accurate for the diagnosis of H. pylori infection. Sampling error is a problem of minor importance. The lower specificity of serological tests may be largely explained by previous treatment of H. pylori.