Fruit and vegetable consumption frequency and tooth-loss transitions across baseline dentition strata in older Chinese adults: a prospective cohort study

BACKGROUND: Fruit and vegetable consumption is linked to oral health in later life, but whether consumption frequency is associated with subsequent tooth-loss transitions across baseline dentition stages remains unclear. We examined these associations in older Chinese adults. METHODS: We used Chinese Longitudinal Healthy Longevity Survey data from 2008, 2011, 2014, and 2018. Participants aged ≥65 years were grouped by baseline dentition; the outcome was incident edentulism in the 1-19-teeth stratum and transition to fewer than 20 teeth in the 20-24 and 25-32-teeth strata. Models mutually adjusted fruit and vegetable consumption frequency. Target-population interpretation prioritised a full baseline-eligible cohort competing-outcome analysis treating death as a competing outcome; additional analyses addressed alternative exposure definitions, differential inclusion, and competing survival. RESULTS: In the full baseline-eligible cohort competing-outcome analysis among participants with 1-19 baseline teeth, more frequent vegetable consumption showed an attenuated inverse association with incident edentulism after accounting for death as a competing outcome (HR 0.88, 95% CI 0.73-1.04). In the observed-outcome sample with follow-up tooth counts, the corresponding vegetable estimate was stronger (HR 0.83, 95% CI 0.69-0.99), whereas fruit consumption was not associated with incident edentulism (HR 0.97, 95% CI 0.86-1.09). Associations in the 20-24 and 25-32 baseline-teeth strata were weaker and less consistent. CONCLUSION: Among older Chinese adults with 1-19 baseline natural teeth, higher vegetable consumption frequency showed only a weak inverse association with incident edentulism when the full baseline-eligible cohort and competing death were considered, whereas fruit consumption remained near null. The stronger estimate from the observed-outcome sample may overstate the association in the target population and should not be interpreted as evidence of an independent protective effect.