The Association of Intake of Vitamin Mixtures With Periodontitis: A Machine Learning Approach on NHANES

Previous studies have focused on the association of single vitamin intake with periodontitis. However, the consumption of vitamins is commonly in the form of mixtures. The relationship between the consumption of multiple vitamins and the prevalence of periodontitis needs to be investigated. In this cross-sectional study, 8512 participants with full-mouth periodontal examination records and vitamin intake (vitamins A, B1, B2, B6, B12, C, D, E, and K) data from the US National Health and Nutrition Examination Survey (NHANES, 2009-2014) datasets were included. GLM and Bayesian kernel machine regression (BKMR) were used to estimate the association of individual and mixed vitamins with periodontitis, respectively. Sensitivity analyses were performed using mean probing depth and mean clinical attachment loss as continuous indicators of periodontitis severity. Vitamins A (OR = 0.90, 95% CI: 0.95-0.99) and E (OR = 0.95, 95% CI: 0.93-0.97) were both negatively associated with periodontitis in GLM estimates and BKMR modeling in the context of vitamin mixtures. Moreover, the intake of vitamin mixtures within a certain range (< 75% percentile) was negatively associated with the estimated risk of periodontitis, which was predominantly driven by vitamins A and E. A potential interaction between vitamins A and E in their association with periodontitis was observed. Similar findings were found in sensitive analyses. BKMR estimates suggested that higher intake of vitamin mixtures might account for decreased odds of periodontitis, with vitamins A and E identified as the most influential factors. Future studies are highly warranted to clarify the underlying mechanisms.