Adjusting estimates of alveolar bone loss for missing observations: developing and testing a general model

The occurrence of missing teeth is a vexing and growing problem in epidemiological studies of dental disease. The number of units of measurement (i.e., teeth/tooth sites) varies between persons and may affect summary statistics in descriptive studies as well as in analytical studies. The purpose of the present study was to develop a general model to adjust summary statistics for missing teeth. The proposed method was applied to alveolar bone loss in current smokers, former smokers, and non-smokers, from 45 to 64 years of age (n = 812). Alveolar bone loss was measured on bitewing radiographs. The adjustment method was based on the assumption that the probability of losing a tooth was an increasing function of alveolar bone loss. The main finding of the present study was that mean alveolar bone loss increased after adjustment for the number of missing teeth. This increase was larger for current smokers than for non-smokers, indicating that the effects of smoking were slightly underestimated when missing sites were ignored. Further research is required for the model to be applied to other data, estimating different types of dental disease with various degrees of disease prevalence and various numbers of missing teeth, and for validating the adjustment method by means of longitudinal data.