Sensitivity and specificity of various tests for the diagnosis of Helicobacter pylori in Egyptian children.

OBJECTIVES

Many noninvasive methods (using breath, blood, and stool samples) are available to diagnose Helicobacter pylori. However, because the noninvasive tests are proxy measures of the infection, they need validation before use. Factors that may affect test validity include patient age, gender, and geographic location. Because no data were available on the validation of noninvasive tests for the diagnosis of H. pylori among children in the Middle East, this study was performed.

METHODS

Children between 2 and 17 years of age evaluated at the Cairo University School of Medicine pediatric gastroenterology clinic who were already scheduled for upper endoscopy were eligible for enrollment in the study. At the time of endoscopy, 3 biopsies were collected and used for rapid urease, histology, and culture, respectively. All children also donated a sample of stool and blood and had a urea breath test performed. Stool and serum samples were tested for the presence of H. pylori by using commercially available enzyme-linked immunosorbent assay-based technology. The sensitivity, specificity, and positive and negative predictive values were calculated for each noninvasive test used in the study. Receiver operating curves also were charted to determine optimal cut points for the various tests when used in the current study cohort.

RESULTS

One hundred eight children were enrolled in the study, with 52 children being under 6 years of age. The urea breath test and HpStar (DakoCytomation, Norden, Denmark) stool enzyme-linked immunosorbent assay kit had the highest sensitivity and specificity (sensitivity and specificity: 98 and 89 [urea breath test] and 94 and 81 [HpStar], respectively), whereas the serologic kit had an unacceptably low sensitivity (50%). The sensitivity of neither the urea breath test nor the HpStar tests was affected by subject age, but specificity of the HpStar test, although still high, was significantly lower among children under 6 years. Receiver operating curves found optimal cut points of the urea breath test at 6.2 delta over baseline and of the HpStar at 0.25 enzyme-linked immunosorbent assay units.

CONCLUSION

The urea breath test and HpSTAR stool antigen kit are reliable tests for the noninvasive diagnosis of H. pylori among children living in the Middle East.