PERIODIC CLUSTERING OF HUMAN DISEASE-SPECIFIC MORTALITY DISTRIBUTIONS BY SHAPE AND TIME POSITION, AND A NEW INTEGER-BASED LAW OF MORTALITY

Human mortality distributions were analyzed for 29 disease-specific causes-of-death in male and female, White (U.S.A.), Black (U.S.A.) and Japanese (Japan) populations, constituting a total of 162 separate cohorts. For each cohort distribution, the curve moments and the parameters values for fits to model equations were determined. The differences between cohort distributions were characterized by two degrees of freedom, related to distribution position and shape, respectively. A form of the Weibull function was shown to contain two parameters that mapped to these two degrees of freedom. Parametric analysis on 136 best-fitting cohorts yielded periodic clustering in the set of values for both Weibull parameters as quantitated using a Fourier transform method and an independent statistical method. This periodicity was unlikely to have occurred by change (P < 0.01). We have combined these results into a Law of Mortality, based on a Weibull function containing only integer parameters and constants, which is valid for all human age-related disease mortality. We show that the life expectancy differences between races and sexes is completely described by this formalism. We conclude that human mortality is controlled by discrete events, which are manifested in the appearance of only allowed mortality curve shapes and positions. © 1990.