[A single year birth cohort analysis of mortality in Japan, 1950 to 1984, statistical method and all cause mortality]

Mortality of single birth year cohort were calculated using the data tape of Information Service on World Health Statistics, WHO, 1950-1984. The data tape gave population and number of deaths for each year of age under 5 years of age and by 5 years of age for ages from 5 to 84 years. Population and number of deaths by age for each year of age in each calendar year were estimated by interpolation. Populations for each year of age were interpolated by B-spline interpolation. Number of deaths for each year of age were calculated by application of "two dimensional semi-Hermite method" after estimation of number of deaths for age 80, 81,......84, using Sprague interpolation factors. Age-specific cohort mortality curves were lower for each successive cohort, demonstrating the so-called "cohort phenomenon", clearly. Ratios of mortalities at the same age, for each successive 5 year interval cohort, which indicate cohort-by-cohort changes of mortalities, were thus less than 1.0 in almost all age groups. In recent years the ratios have gradually been leveling off at 0.8-0.9. Cohorts born in the early Showa Era (around 1930) showed higher ratios than all other cohorts, indicating relatively poor improvement in mortalities compared to the other cohorts. Ratios of cohort mortalities for successive ages within the same cohort, which indicates mortalities increase with advance of age, were greater than 1.0 after adulthood. These particular results showed that the increase in mortality with age accelerated after 30 years of age and that a difference in trend was observed between male and female after age 50. This new method of cohort analysis, single birth year cohort, gives clearer results than those obtained by methods heretofore utilized in cohort studies and is useful for more detailed analysis of cohort mortality.