Tests for Helicobacter pylori infection:: a critical appraisal from primary care

Background. Testing of patients for Helicobacter pylori infection is common in primary care settings. The accuracy of such tests has been assessed and critical appraisal of this literature can inform the clinical management of patients suspected of being infected with H.pylori. Methods. Literature evaluating the accuracy of diagnostic tests for H.pylori infection was sought as part of a systematic review of literature concerning the management of patients infected with H.pylori. Studies were appraised and estimates of sensitivity and specificity were extracted. Positive and negative likelihood ratios (LRs) were calculated and the implications for post-test probabilities are reported. Results. The sensitivity, specificity, LR+ and LR- for H.pylori infection tests are: [C-13]urea breath test (UBT), 96.5, 96, 24 and 0.04; [C-14]UBT, 97.5, 95.5, 21 and 0.03; serology, 91, 89.5, 8 and 0.11; near patient tests, 77, 74, 3 and 0.31; and meta-analysis of serology, 85, 79, 4 and 0.19. The range of pre-test probabilities of H.pylori infection in which the diagnostic tests were useful, estimated from primary studies, were: [C-13]UBT, 20-90%; [C-14]UBT, 20-99%; serology, 30-80%; and near patient tests, 50-60%. Conclusions. Tests for H.pylori infection are useful in primary care when the pre-test probability of infection is neither too high nor too low. This indicates that the tests may not be useful for screening purposes but may help with differential diagnosis. Outside moderate pre-test probability ranges, the chances of a result being false is high, and such patients should either receive eradication without prior testing (if the probability of infection is sufficiently high) or the test result should be reconfirmed. When the pre-test probability falls below similar to 20%, a positive test result is unreliable. If the pre-test probability is above similar to 80%, a negative test result is unreliable. Clinical selection of patients needing testing should be used to limit testing to individuals with pre-test probabilities within these ranges. The choice of diagnostic H.pylori test should be influenced by the H.pylori infection rate in the population being tested and the test characteristics. Recommendations for the use of tests, especially near patient tests, should be reconsidered. This critical appraisal supports the recommendations of the European Society for Primary Care Gastroenterology guidelines, arrived at by consensus, for testing for H.pylori infection in primary care.