Dynamics of human mortality.

Historical human mortality curves display five phases, differing in dimensions with population, time and circumstance. Existing explanatory models describe some but not all of these, and modelling of entire curves has hitherto necessitated an assumption of multiple distributions. A new distribution, shown previously to describe survival in experimental animals, postulates empirically that net mortality risk comprises two components described as 'redundancy decay' and 'interactive risk'. The former is proposed effectively to set a 'program' for the increase of mortality risk with age, and may be a better measure of the 'rate of ageing' which has previously been assumed to determine the slope of semi-logarithmic mortality curves. Entire human mortality curves are shown here to be compatible with this single distribution given the assumption that individuals vary only with respect to the interactive risk parameter (k). Historical Swedish cohort data are modelled here exclusively through changes in k values, clustered increasingly toward the higher end of their range, while redundancy values are held constant. This pattern is compatible with the hypothesis that historical changes in human mortality may be explained purely in terms of interactive risks and without changes in the underlying pattern or rate of ageing.