Self-rated oral health over time and incident stroke in later life: a longitudinal cohort study

Poor self-rated oral health has been linked to stroke, but evidence beyond single baseline measures remains limited. We examined whether short-term and long-term patterns of self-rated oral health were associated with incident stroke and whether systemic inflammation accounted for part of these associations. We used data from the English Longitudinal Study of Ageing. Oral health was assessed using a single self-rated item concerning the health of the mouth, teeth and/or dentures. Wave 7 was baseline for short-term oral health and covariates, and self-reported physician-diagnosed incident stroke was ascertained through Wave 10. Oral health at Waves 3, 5 and 7 was used to derive long-term burden, repeated poor oral health and trajectories. Associations were examined using multivariable Cox proportional hazards models, and exploratory mediation analyses evaluated whether C-reactive protein (CRP) and fibrinogen measured at Wave 8 accounted for part of these associations. The short-term and long-term analyses included 7,886 and 4,750 participants, with 495 and 332 incident strokes reported over a mean follow-up of approximately 5.9 years, respectively. Better short-term oral health was associated with lower stroke risk (per 1-point higher score: HR 0.86, 95% CI 0.79–0.93), as was better long-term oral health (HR 0.76, 95% CI 0.66–0.87; highest vs. lowest tertile: HR 0.65, 95% CI 0.49–0.86). Poor oral health at all three assessments was associated with higher stroke risk than no poor oral health (HR 2.03, 95% CI 1.39–2.99). More favourable trajectories were associated with lower risk. Exploratory mediation analyses suggested small and imprecisely estimated indirect effects through CRP and fibrinogen. Poorer self-rated oral health was associated with higher stroke risk across short-term and long-term measures. The clearest excess risk was observed among participants with repeatedly poor oral health, whereas systemic inflammation explained only a small proportion of the association.