The effect of measurement error on the determination of Helicobacter pylori prevalence.

We assessed the effect of serum-based antibody tests on the epidemiology of Helicobacter pylori infection. We took crude population prevalences of H. Pylori infection from existing publications in which antibody-based tests were used to determine prevalence. We then calculated 95% confidence intervals that included terms for study size, sensitivity, specificity, and, where possible, the sample size used to determine the validity of the antibody test. Specificity had a greater effect than sensitivity on the overestimation of most population-based estimates of H. pylori prevalence. The attributes of some antibody-based tests imply that no matter how large the study size, an accurate estimate of prevalence could not have been obtained.