From Mouth-level to Tooth-level DMFS: Conceptualizing a Theoretical Framework

There is no dearth of correlated count data in any biological or clinical settings, and the ability to accurately analyze and interpret such data remains an exciting area of research. In oral health epidemiology, the Decayed, Missing, Filled (DMF) index has been continuously used for over 70 years as the key measure to quantify caries experience. The DMF index projects a subject's caries status using either the DMF(T), the total number of DMF teeth, or the DMF(S), counting the total DMF teeth surfaces, for that subject. However, surfaces within a particular tooth or a subject constitute clustered data, and the DMFS mostly overlook this clustering effect to attain an over-simplified summary index, ignoring the true tooth-level caries status. Besides, the DMFT/DMFS might exhibit excess of some specific counts (say, zeroes representing the set of relatively disease-free carious state), or can exhibit overdispersion, and accounting for the excess responses or overdispersion remains a key component is selecting the appropriate modeling strategy.

This concept paper presents the rationale and the theoretical framework which a dental researcher might consider at the onset in order to choose a plausible statistical model for tooth-level DMFS. Various nuances related to model fitting, selection and parameter interpretation are also explained.

The author recommends conceptualizing the correct stochastic framework should serve as the guiding force to the dental researcher's never-ending goal of assessing complex covariate-response relationships efficiently.