ESOC 2025 | An analysis of the ARIC study that adjusts for potential confounding bias

So this is again from the ARIC study. Okay. Previously at the Stroke Conference in the US, we presented data that oral flossing reduces risk of primary stroke, primary incident ischemic stroke. However, a question is always raised whether those who floss also have other good healthy habits. And do they also have lower stroke risk factors? So there might be some confounding bias. So in order to reduce that confounding bias, what we did was we actually created a propensity score-based model. So we used the things that might cause confounding bias, such as the risk factors and other oral health behaviors to generate a propensity score. And then we matched within that ARIC data set the patients who flossed and did not floss. So based on this, we were able to create a data set of about 1,700 patients who flossed, subjects who flossed, and 1,700 subjects who did not floss. They were all propensity score matched. And then we again used the follow-up data for 25 years, and it shows that those who flossed regularly in the propensity match data set had a lower rate of stroke. What it shows is that those who flossed had more than a 20% risk reduction, similar to what was shown in the original study, except that this one took out propensity score, but to use propensity score matching to reduce that confounding bias. Further, because flossing frequency is associated with ischemic stroke, we ran a mediation analysis which showed that flossing frequency significantly lowered the rate of combined gum disease and tooth decay, and the reduction in gum disease and tooth decay resulted in reduced inflammation, C-reactive protein and IL-6, and that resulted in reduction in ischemic stroke. So this adds two things. One, it removes confounding bias and also shows how flossing might be related to reduction in ischemic stroke. And again, that’s a 20% risk reduction, but mediated by oral health conditions such as gum disease and tooth decay mediated by lowering of inflammation measured by C-reactive protein and interleukin 6.

This transcript is AI-generated. While we strive for accuracy, please verify this copy with the video.