On the statistical evaluation of adherence assays.

Parametric (unpaired t-test) and non-parametric (Mann-Whitney U-test) methods have been used in the evaluation of adherence assays on the non-antibiotic antimicrobial agent, Taurolin. In all but one case, where the anti-adherence effect was known to be marginal, both statistical methods gave similar results although there were some minor differences in the levels of significance achieved. The effect of the agent on the deviation of adherence data from normality was quantified by calculation of the skewness coefficient for each data set. A significant anti-adherence effect appears to result in a decrease in the skewness of the adherence assay data. It was concluded that either parametric or non-parametric statistical evaluation of adherence assay data is valid for large numbers of observations. In future studies of this type it is suggested that attention should also be given to the effect of the anti-adherence agent on the deviation of adherence data from normality as denoted by the skewness coefficient.