Analysis of covariance in caries clinical trials using groupings of teeth

An empirical investigation was carried out on data from four caries clinical trials of subjects who were initially 11 to 12 years old. The three baseline variables suggested by Adkins (1977), as possibly being related to caries increment, were each calculated on specific groupings of teeth, and these were then considered as predictor variables for the dependent variable, caries increment. A gain in efficiency of approximately 20% was obtained when an analysis of covariance was used with these predictor variables, over a straightforward analysis of variance. However, this was only slightly more efficient than using baseline DMFS as a covariable. This investigation did not support the results of initial work done by Adkins (1977).