A lifestyle-derived risk score for chronic atrophic gastritis: development and validation in 21,008 Chinese adults

Background: Chronic atrophic gastritis (AG) is a key precancerous lesion of gastric cancer. Modifiable lifestyle factors such as smoking, alcohol consumption, and specific dietary patterns are associated with the risk of AG. However, lifestyle indicators have not been integrated into clinical practice, and there is a lack of practical integrated risk assessment tools for primary healthcare settings. Objective: Develop and validate a simple lifestyle risk score to identify individuals at high risk of AG in the Chinese screening population. Methods: This population-based case-control study utilized data from the Upper Gastrointestinal Cancer Screening Program in Gansu Province, China. A total of 21,008 participants aged 40-70 years were included. AG status was determined based on endoscopic and pathological diagnoses. A risk score (range 0-7) was developed incorporating seven modifiable lifestyle factors, with fried food and garlic intake included as exploratory components based on prior literature. Associations were assessed using multivariate logistic regression, and predictive performance was evaluated through ROC curve analysis, calibration, and decision curve analysis. Results: for trend <0.001). After adjusting for age and sex, each one-point increase in the score was associated with a 51% increase in AG risk (adjusted OR = 1.506, 95% CI: 1.460-1.554). Compared with the low-risk group (score 0-1), the high-risk group (score ≥3) showed a significantly elevated AG risk (aOR = 3.614, 95% CI: 3.289-3.971). The area under the ROC curve for the score was 0.597 (95% CI: 0.590-0.604). The model demonstrated good calibration, and decision curve analysis indicated that applying this score in clinical decision-making provided net clinical benefit across a wide range of threshold probabilities (approximately 10-60%). Conclusion: This study developed and validated a simple lifestyle risk score that demonstrated a significant association with AG risk and modest but clinically useful predictive performance suitable for primary healthcare settings. This tool is applicable for rapid identification of high-risk AG individuals in primary healthcare settings and gastric cancer screening programs, providing practical evidence to guide targeted preventive interventions.