A Pragmatic Risk Index Evaluating the Elderly with Comorbidity for Oral Health Event Times

Abstract With the rising cost of dental healthcare spending, future professional dental treatment plans will seek to utilize efficient prognosis to weigh the long-term outlook of tooth retention against the costs involved. This is in context when tooth loss (due to caries, or periodontal disease, or both) continues to remain a major oral health burden in the US, while the Affordable Care Act, although covering the pediatric population, is not mandated for the adults. Furthermore, elderly people with additional comorbidities, such as Type-2 diabetes (T2D), are often at an elevated risk of compromised oral health. With recent significant advances in data storage capabilities, huge oral health epidemiologic databases, such as those generated from the ‘AxiUm’ management system, or national surveys containing a wealth of risk information and associated oral health time to event (TTE) information are at our disposal. However, their comprehensive exploration has mostly gone untapped, mainly due to the unique statistical complexities they pose, that are often beyond the capabilities of existing statistical tools and software packages. These include, but not limited to (a) an extremely multi-level setup, (b) very high level of censoring patterns, (c) issues related to informative sampling in national surveys, and (d) no consensus on the ‘effective’ and ‘clinically interpretable’ summary of the complex effects of predictors, and their confounders, on these TTE endpoints. The goals of this project are to address these complexities and perform a comprehensive statistical analysis of available TTE endpoints from 3 rich databases comprising the U.S. Midwest and southeast population, and a nationally-representative NHANES (2009-2014) database consisting of about 11,700 adult dentate subjects. Our proposal has three broad aims of practical interest to oral epidemiologists and practitioners: (a) construct a pragmatic risk index for estimating the median residual lifetime of a tooth for the T2D/Non-T2D elderly, (b) refine that index to estimate a tooth’s time to landmark events in a nationally-representative complex survey, and (c) to make these newly developed statistical and epidemiological tools and results widely accessible (for free) to clinical practitioners. Specifically, we propose to develop a well-documented, free, web-application, where a user/clinician can plug-in values of a subject’s characteristics to calculate the corresponding index determining tooth TTEs. This proposal is expected to generate new knowledge on evaluating national-level risk assessments of tooth survival via this comprehensive and unique index. To the best of our knowledge, this idea of creating a tooth TTE index is the first of its kind, and will provide a set of data analytic tools that are readily generalizable. When nationally- representative longitudinal databases are made available (from the ongoing NHANES studies), the proposed methods can be readily adapted to inform interventions and treatment decisions.