Simulation experiments of tumor measurement in clinical trials

A simple model of tumor growth involving stochastic growth factors as well as measurement errors in explored in the context of a simulated "clinical trial" comparing two treatments. The model shows how treatment effects on tumor growth translate themselves into "improvement rates" derived from measurement on diameters. Furthermore, it shows that treatment contrasts of considerable magnitude (such as doubling of mass or volume versus shrinkage to one half) may escape recognition in clinical trials. A number of "power" tables are presented that provide the probability of making correct judgments of treatment effects. The tables indicate that the actual power in trials of this nature is in satisfactory agreement with that claimed in the standard power tables which are currently used in guiding trial strategy based on improvement rates.