Confidence limits for low induced frequencies of oncogenic transformation in the presence of a background.

The problem is addressed of assessing realistic confidence limits for oncogenic transformation rates when the incidence of transformation is small, and there is a background present. It is shown that the commonly used approximations for assessing the confidence interval on the radiation-induced transformation rate (t1) and the control rate (t2) are inadequate when only a small number of transformants are observed. Accurate techniques are described for these cases for the estimation of confidence limits of t1, t2 and also t1-t2.