Sample Size Considerations in Designing Studies with Intra-oral Models

Especially during recent years, the use of pre-clinical models for predicting the efficacy of fluoride systems has assumed greater importance within the scientific community. Originally utilized primarily to screen experimental fluoride delivery systems, preclinical models are now being considered as predictors of clinical efficacy in lieu of controlled clinical caries trials. Of the various preclinical models presently available, human intra-oral models have the greatest potential for reflecting intended usage conditions and therefore may be the most meaningful models for predicting clinical efficacy. However, only with the proper consideration of numerous critical variables can studies using intra-oral models be appropriately designed to achieve the desired objectives. Clearly, these models must provide relevant information in a manner which reflects clinically established cariostatic activity and be capable of detecting established differences in the amount of cariostatic activity, i.e., dose-response effects. Three sources of variation must be considered before an appropriate study design and sample size can be chosen. Based on fluoride uptake data from an intra-oral model with proximally-located enamel specimens, estimates of variation among subjects, within subjects, and among specimens within subjects were obtained. Multiple specimens per panelist do not affect the first two sources of variation. Thus, the number of panelists, and not the number of specimens, is of primary importance when pre-test data are used to choose the appropriate study design and calculate the required sample size.