Examining the Relationship of Periodontal Treatment and Improved Health Outcomes for Coronary Heart Disease, Cerebrovascular Disease and Diabetes

项目摘要 有限的证据表明，牙周治疗可以改善心血管和糖尿病的结果， 一些团体。尽管有这些发现，但我们对以下关系的理解仍然存在重大差距： 牙周病和这些医疗条件。拟议的研究将利用一个关系数据库 结合12年的医疗，药房，实验室，牙科和保险索赔数据， 以社区为基础的综合卫生系统，探讨牙周病的治疗是否可以改善 在控制以前未解决的混杂因素时，选择医疗结局并降低医疗成本 使用观察数据的研究。通过利用纵向（重复测量）数据，我们将获得 与以前的报告相比，在更长的后续行动时间内获得更相关的措施。此外，我们将 使用额外的统计方法-与传统的流行病学结果并排比较-通过 设计，旨在控制未测量的混杂因素（例如，态度、个性、未观察到的健康习惯）， 随着时间的推移保持相对稳定或旨在利用数据中的准实验性质（即， 工具变量）来近似因果效应，就好像牙周治疗是随机分配的一样。我们 预期效果估计相对于先前报告的结果可能会减弱（甚至消除 当通过使用更丰富的数据和更复杂的分析方法更好地控制混杂时， 这些更丰富的数据。如果以前的发现能经得起我们更有力的探测，我们相信， 对这些结果的信心将为临床优先事项的确定带来实质性价值。

Project Summary Limited evidence suggests that periodontal treatment may improve cardiovascular and diabetes outcomes in some groups. Despite these findings, critical gaps remain in our understanding of the relationship between periodontal disease and these medical conditions. The proposed study will utilize a relational database combining 12 years of medical, pharmacy, laboratory, dental, and insurance claims data from a large, community-based, integrated health system to explore whether treatment of periodontal disease can improve select medical outcomes and reduce medical costs when controlling for confounders not previously addressed in studies utilizing observational data. By leveraging longitudinal (repeated measures) data, we will have access to more relevant measures over longer follow-up times relative to previous reports. In addition, we will use additional statistical approaches—compared side by side with traditional epidemiological results—that, by design, aim to control for unmeasured confounders (e.g., attitudes, personality, unobserved health habits) that remain relatively stable over time or aim to leverage quasi-experimental properties in the data (i.e., instrumental variables) to approximate causal effects as if periodontal treatment were randomly assigned. We anticipate that effect estimates may be attenuated (or even eliminated) relative to previously reported results when confounding is better controlled for by using richer data and more sophisticated analytic methods afforded by those richer data. If previous findings hold up to our more robust probing, we believe the added confidence in those results will bring substantial value to informing clinical priority setting.