Diagnostic Performance for Detection of Helicobacter Pylori Infection in Gastric Biopsy Specimens with No Gold Test: Non-linear Regression Approach

초록

The selection of a test as a reference with no perfect sensitivity and specificity may lead to bias, yielding distortion of the diagnostic performance. This means it is inappropriate to use imperfect diagnostic tests as a reference method to identify infected patients in clinical environments. In this study, diagnostic performance of rapid urease test, polymerase chain reaction (PCR), and histology of gastric biopsy specimens for diagnosing Helicobacter pylori infection separately and in combination was estimated by using non-linear regression. Based on this approach, the sensitivity, specificity and likelihood ratio positive and negative values for each test were as follows: urease test 99.9%, 99.9%, 99.9%, 99.6%, respectively; PCR 88.6%, 99.9%, 99.9%, 70.5%, respectively; histology 78.3%, 97%, 78.3%, 97%, respectively. Predictive values for positive and negative changes with varying prevalences. A positive histology test will be correct at least 90% of the time provided prevalence of H. pylori is greater than 35%, whereas a negative test will be correct in more than 90% if the prevalence is less than 25%. Combination of three diagnostic tests employed in the study gives no substantial benefit for practitioners to screen infected patients, and urease test or PCR represents an appropriate single test in clinical environments.

키워드