The overall goal of this proposal is to undertake an integrative statistical examination of data obtained from the Iowa Fluoride Study. In the United States, dental caries is a major chronic childhood disease. Nevertheless, our understanding of various risk factors is limited. Novel statistical models for temporal and statistically correlated count data with excessive zeros will be developed for analyzing tooth level data using a Bayesian formulation. The Iowa Fluoride Study (IFS) is an ongoing study of a cohort of Iowa children that began in 1991, led by Dr. Steven Levy who is a co-I on this proposal. As such, the resulting dataset is a potential source of valuable information. It is anticipated that, through innovative and efficient statistical modeling, it will be possible to mine these data more fully and discover novel relationships between caries incidences/severity and various potential risk and preventive factors and study their short- and long-term effects. We will also look at the data on dental fluorosis that were collected on the same subjects. A different statistical model for ordinal temporal data will be developed for this purpose. It is anticipated that, overall, we will be able to determine the preventive factors for dental caries which are not detrimental from the standpoint of fluorosis development. Thus, the following two broad interconnected aims will be undertaken. We will develop count data models to study the dynamic changes of the various factors on caries severity including efficient computational methods for Bayesian inference (Aim 1). These models will be able to integrate the entire tooth level data on caries experience scores for all children in the IFS for ages 5, 9, 13 and 17 years by introducing sources of spatial and temporal correlations. We will also develop temporal ordinal data regression models to study the tooth level dental fluorosis scores at 9, 13 and 17 years (Aim 2). Once again, algorithms for efficient Bayesian computation will be developed. We will compare our results to those obtained from existing approaches and also results available in the existing caries literature. Statistical software (R packages/codes) implementing the temporal clustered count data analysis methods will be freely distributed through the Comprehensive R Archive Network.
The proposed research will lead to a detailed statistical examination of caries and fluorosis data obtained from the Iowa Fluoride Study. This comprehensive research plan will integrate multi-year caries and fluorosis data to uncover new information that are undetectable using the simpler approaches previously undertaken for the IFS data. This research provide a greater understanding of the risk and preventative factors of dental caries and fluorosis. Last but not least, this work will enrich biostatistics methodological research in statistical regression models for count data.