Accuracy of diagnostic methods used for epidemiological studies of Helicobacter pylori.

Epidemiological studies involve groups of individuals, or whole populations, many or most of whom are not ill. Clinical investigation has an individual perspective and precise statements need to be made about the individual alone. Serological methods are most commonly used for population-based epidemiological studies. Non-invasive epidemiological methods, using breath tests or the study of saliva or urine, are increasing in use. All methods depend on accuracy in identifying presently infected or non-infected persons, and accuracy in defining a previously infected person. The performance of serological methods varies with the antigens chosen, the population from which reference sera are drawn, age, ethnicity, and homologous and heterologous infection rates in the population being studied. Much of the standardization of epidemiological assays has been done in adults, which means that for children there is still uncertainty concerning standards and cut-off values. Because of differences in strains of H. pylori in different geographical areas, antigen selection is important when geographical comparisons are made. The sensitivity and specificity of a test is not strongly affected by the prevalence of infection. However, as the prevalence rises in the tested populations, the reported positive predictive value rises, and the negative predictive value falls. Depending on the patient population studied, accuracy varies with changes in the prevalence, and its magnitude depends both on the sensitivity and specificity. Accuracy is therefore not a very useful measure. It is better to look at the sensitivity and specificity, and the prevalence in the study, where they are measured. An alternative to separating test results into two or three categories is to report likelihood ratios, which report the probability of a person with a particular result being truly positive compared with the probability of a person with that result being truly negative. A receiver operating characteristic (ROC) curve, which describes the effect of varying the cut-off value on the performance of a test, can be useful in comparing the performance of two or more different tests. The use of multiple tests to augment positive culture as a 'gold standard', has been aided by use of polymerase chain reactions and other molecular biological methods. However, this augmentation has its limitations, since each of the additional methods may produce false positives. For example, polymerase chain reactions can be falsely positive if instruments are contaminated. There are people with more than one strain of Helicobacter pylori in the stomach, and, without molecular biological efforts, or serological typing tests, the 'gold standard' does not deal with multiple infections.