Abstract
Using Japanese Vital Statistics, the average (five years) annual mortality rates from cerebrovascular diseases (B22 in the 7th revision of ICD and B30 in the 8th revision of ICD) were calculated for the years 1960, 1965, 1970, and 1975, for males and females ranging in age from 35 to 74, grouped into five-year age groups. Two mathematical models rij=cαiβ'jγk (Model A) and rij=cαiγkexp (βjf⋅δl) (Model B) were postulated, and their fit to the observed data was tested. Where, rij is the population mortality rate in cell (i, j), c is a constant, αi, βj, γk represents the effect of ith age group, jth year of death and kth birth cohort, respectively, and exp (βiδl) represents the effect of jth year of death in the lth age group.
For male data, Model B fit very well to the observed data, but the fit of Model A was not good. For female data, both Model A and Model B fit relatively well to the observed data, but the fit of Model B was better than the fit of Model A. Model B seems to be a mathematical model worthy of further consideration, although some estimates of parameteres in Model B throw slight suspicion on the reality of this model.
