To assess the concordance and performance characteristics of Helicobacter pylori laboratory tests compared with histopathology and to propose algorithms for the diagnosis of H pylori that minimize diagnostic error.
H pylori diagnostics were reviewed from a 12-year period within a health system (2,560 cases). Analyses were performed to adjust diagnostic performance based on treatment and consensus histopathologic diagnoses among pathologists. Markers of access to care, including test cancellation frequency and turnaround time, were assessed. Costs and performance of candidate noninvasive testing algorithms were modeled as a function of disease prevalence.
Serum H pylori IgG demonstrated a higher sensitivity (0.94) than urea breath and stool antigen tests (0.64 and 0.61, respectively). Evidence of an advantage in access to care for serology included a lower cancellation rate. Interobserver variability was higher (κ = 0.34) among pathologists for cases with a discordant laboratory test than concordant cases (κ = 0.56). A model testing algorithm utilizing serology for first-time diagnoses minimizes diagnostic error.
Although H pylori serology has modestly lower specificity than other noninvasive tests, the superior sensitivity and negative predictive value in our population support its use as a noninvasive test to rule out H pylori infection. Reflexive testing with positive serology followed by either stool antigen or urea breath test may optimize diagnostic accuracy in low-prevalence populations.