Xerogenic medications and clinical markers of salivary hypofunction.

OBJECTIVES: To determine whether xerogenic medication-class exposure and chair-side salivary markers contribute to risk stratification of objective salivary hypofunction across a broad clinical age spectrum, using conventional regression and explainable machine-learning approaches. MATERIALS AND METHODS: This retrospective cross-sectional study included 850 patients (mean age, 60.8 ± 15.6 years; 657 [77.3%] female) who underwent salivary flow assessment and oral clinical evaluation at Kyung Hee University Dental Hospital. Patients were categorized into four age groups: <40, 40-64, 65-79, and ≥ 80 years. Unstimulated salivary flow rate (UFR), stimulated salivary flow rate (SFR), and flow-rate-defined hyposalivation were analyzed using age-group comparisons, multivariable regression models, incremental gradient boosting models, and SHapley Additive exPlanations (SHAP)-based model interpretation. RESULTS: Mean UFR was significantly lower in older age groups, ranging from 0.57 ± 0.30 mL/min in patients aged < 40 years to 0.36 ± 0.24 mL/min in those aged ≥ 80 years (p < 0.001), whereas age-group differences in SFR were less pronounced. SFR-defined hyposalivation was most frequent in the ≥ 80-year group (17.3%; p = 0.007). In adjusted models, the ≥ 80-year group had lower UFR than the < 40-year group (β = -0.194; 95% CI, - 0.272 to - 0.115; p < 0.001) and higher odds of SFR-defined hyposalivation (OR = 5.47; 95% CI, 1.44-20.78; p = 0.013). Female sex was associated with lower UFR (β = -0.055; p = 0.009) and SFR (β = -0.369; p < 0.001). High-confidence xerogenic medication-class count was independently associated with SFR-defined hyposalivation (OR = 1.63; 95% CI, 1.12-2.36; p = 0.010). In incremental machine-learning models, adding salivary pH and buffer capacity produced the largest improvement in discrimination. AUROC increased from 0.468 to 0.723 for UFR-defined hyposalivation (ΔAUROC = 0.255; DeLong p < 0.001) and from 0.608 to 0.725 for SFR-defined hyposalivation (ΔAUROC = 0.116; DeLong p = 0.005). SHAP analysis ranked salivary buffer capacity as the leading contributor for both outcomes, with salivary pH also among the top predictors. CONCLUSIONS: Objective salivary hypofunction was not fully explained by age, systemic disease burden, or medication burden alone. Instead, chair-side salivary functional markers, particularly salivary pH and buffer capacity, captured clinically relevant information about current salivary status. These findings support a practical, layered approach to identifying patients who may require objective salivary flow assessment. CLINICAL RELEVANCE: Chair-side salivary pH and buffer capacity may help clinicians identify patients who warrant objective salivary flow assessment, particularly when considered together with age, sex, and xerogenic medication-class exposure.