Dental fear and oral health outcomes among college students: a cross-sectional study with machine learning interpretability analysis

Dental fear and anxiety (DFA) is a recognized barrier that prevents people from seeking dental care. Although DFA has been examined across various age groups, limited evidence exists on how multidimensional dental fear relates to objective oral health indicators among college students. This cross-sectional study examined how different aspects of dental fear are associated with oral health in Chinese university students. Using a stratified cluster sampling approach, we surveyed 490 students at Yanbian University in China. Dental fear was measured with the Chinese version of Stouthard’s Dental Anxiety Inventory (DAI), which covers five dimensions: fear of treatment procedures, anxiety during dental visits, general dental anxiety, fear of the dentist’s opinions, and fear of the overall visit experience. Standardized clinical examinations recorded the Decayed, Missing, and Filled Teeth (DMFT) index, calculus deposits, and gingival bleeding. Group comparisons used t-tests and one-way ANOVA, while multiple linear regression quantified associations between each fear dimension and oral health indicators. A gradient boosting machine (GBM) combined with SHapley Additive ExPlanations (SHAP) analysis provided interpretable insights into feature contributions and dose-response patterns. Participants had an average total dental fear score of 88.84 ± 38.56 (on the 36–180 DAI scale; converted to 2.47 ± 1.07 on a 5-point scale). Caries were found in 77.96% of students, with a mean DMFT of 3.62. Anxiety during the dental visit process was the only dimension that differed significantly by gender (t = − 3.74, p < 0.01) and by treatment experience (t = 3.29, p < 0.01). In the full regression models, anxiety during dental visits (β = 0.524, p < 0.01), general dental anxiety (β = 0.689, p < 0.05), and fear of the dental visit process (β = 0.504, p < 0.05) were significantly associated with higher DMFT. Fear of the dentist’s comments was the only dimension significantly associated with calculus (β = 1.233, p < 0.001). None of the fear dimensions were significantly associated with gingival bleeding after adjusting for covariates. SHAP analysis confirmed DMFT and calculus as the leading oral health predictors of fear intensity, with a clear dose-response pattern emerging above fear scores of 4.0. Dental fear—particularly the general anxiety dimension and anticipatory anxiety about the visit process—was associated with higher DMFT and more calculus deposits in college students. No significant associations were found between any fear dimension and gingival bleeding after adjusting for covariates. We found a monotonic positive relationship between reported fear levels and predicted DMFT and calculus values, with the effect becoming particularly strong at high fear scores. These results suggest that university dental services should include fear screening and targeted interventions, especially for students reporting high general dental anxiety.