ANALYTIC APPROACHES TO LONGITUDINAL CARIES DATA IN ADULTS

The objective of this paper is to consider current methods for analyzi ng longitudinal caries data in adults. To illustrate these methods, we used data from the Piedmont dental study, a prospective investigation of the oral health of older adults. Longitudinal dental data sets com prise repeated observations of an outcome (often clustered within rand omly selected primary sampling units), and a set of covariates for eac h of many subjects, in whom clustering can occur as a result of measur ing teeth, or surfaces, within people. One objective of statistical an alysis is to predict the outcome variable as a function of the covaria tes, while accounting for the correlation among the repeated observati ons for a given subject and the effect of clustering within subjects, as well as between subjects within primary sampling units, such as com munities, schools, hospitals, or other such units. We considered two s tatistical approaches: generalized estimating equations and survey reg ression models. We also examined the impact of varying diagnostic crit eria for caries estimation between epidemiologists and clinicians. One approach is to perform the usual time(x) exam score minus time(0) sco re analysis for the baseline and final examinations, while an alternat ive is to analyze trends among interim examinations. Finally, because caries studies in which the onset of the disease is the endpoint face the problem of censoring due to subject attrition and/or tooth loss, w e recommend the incidence density (time-to-event) analytic strategy to address this problem. This approach was found to be most suitable for longitudinal studies of older adults since it accounts for the time e ach surface remains at risk for the event of interest, making use of i nterim exam data until the moment the subject and/or the tooth are no longer available for examination. We also included a discussion on bia ses that occur upon application of the usual methods of estimating car ies experience in missing teeth and crowns, which often ignore the cla ssification error in the estimation. We propose a method to adjust for misclassification of the hi-component of the DMFS index. In the case where one can observe true reversals or remineralization of caries les ions, we recommend an adjustment formula to account for reversals that are most likely due to examiner misclassification. We provide example s to demonstrate the applicability of the methods for covariates subje ct to outcome misclassification.