Abstract
BACKGROUND: Dyspepsia is a common indication for gastroscopy, yet its diagnostic yield for malignancy and other clinically significant findings (CSF) remains low. Improved risk stratification is therefore essential to guide endoscopic referral. This study evaluates the diagnostic yield of gastroscopy in dyspepsia and investigates the predictive roles of age, ethnicity, and alarm symptoms. METHODS: This retrospective single-center study was conducted at a university-affiliated hospital in Israel and included 3022 patients who underwent gastroscopy for dyspepsia over a five-year period. Multivariate logistic regression identified independent predictors of CSF, and receiver operating characteristic (ROC) analysis determined optimal age thresholds for malignancy and CSF. RESULTS: Functional dyspepsia accounted for 55.9% of cases, while precancerous gastric lesions and upper gastrointestinal malignancies were identified in 12.8% and 0.79%, respectively. In multivariable models, age ≥ 50 years (OR = 2.59; CI: 2.02-3.32) and alarm symptoms (OR = 1.79; 95% CI: 1.33-2.41) independently predicted CSFs. Malignancy was similarly associated with age ≥ 50 years (OR = 4.89; CI: 1.11-21.60) and alarm symptoms (OR = 31.42; CI: 10.26-96.19). ROC analysis identified optimal age thresholds of 50 years for CSF (AUC = 0.65) and 54 years for malignancy (AUC = 0.72). Ethnicity did not independently predict malignancy, though minority patients showed differing precancerous lesion patterns. CONCLUSIONS: Age ≥ 50 years and alarm symptoms significantly increased the likelihood of CSFs and malignancy, supporting a selective approach to gastroscopy. ROC-derived thresholds may support reconsideration of age criteria in settings with similar epidemiologic patterns, highlighting the need for region-specific risk stratification.