The lognormal frequency distribution and human cancer

Abstract

Many observations or measurements associated with the natural history of human cancer, including growth rates, mass, and survival, tend to form skewed frequency distributions that often approximate lognormality. The random variation of events relating to human cancer is often geometric or multiplicative rather than arithmetic or additive. Plotting these skewed distributions with the parameters of lognormal distributions permits an accurate description of the geometric mean and a more efficient calculation of the variance. Extreme variance in behavior now has a proper mathematical description, and other statistical tests can be performed with greater accuracy using the logarithmic parameters that better describe the frequency distributions of measurements or observations under consideration.