Can a Simple Clinical Score Predict H. pylori Infection Before Testing?: Frontiers in Medicine | September 2026

What Did the Study Find?

This study developed a clinical prediction model for Helicobacter pylori infection using 2,034 participants from three centres in Qinghai Province, China, with infection confirmed by ¹³C-urea breath testing.

• H. pylori prevalence was approximately 52%.
• Seven readily available predictors were retained: educational level, vegetable intake, tooth-brushing frequency, halitosis, family history of H. pylori, chronic gastritis and high-altitude residence.
• The model achieved an AUC of 0.774 in training and 0.756 in internal validation.
• Calibration was good, and decision-curve analysis suggested potential clinical utility.
• The model was converted into a nomogram for individualised risk estimation.
• Importantly, “high altitude” was a region-level proxy, not an individually measured exposure.

Why Does It Matter?

A questionnaire-based tool could potentially help prioritise H. pylori testing where breath tests or stool-antigen testing are difficult to provide universally.

GastroAGI Takeaway

This model predicts who should be tested—it does not diagnose H. pylori. Its moderate discrimination may be useful for targeted screening in resource-limited or geographically remote populations, but it should not replace validated diagnostic testing. The cross-sectional design, internal-only validation and region-specific predictors mean external prospective validation is essential before wider clinical use.