[An analysis of disparities in the changes of cancer mortality rates among prefectures in Japan using age-period-cohort analysis].

Objectives　In this study, we compared the decrease in cancer mortality rates among prefectures in Japan using age-period-cohort (APC) analysis.Methods　We used the cancer mortality data of each prefecture in Japan, as determined by the Vital Statistics, over 5-year periods from 1995 to 2015. Records of the number of mortalities in each 5-year age group from 40-44 to 85-89 years age was collected. We fitted a Bayesian APC model to the data of each prefecture and estimated the birth cohort effect on cancer mortality rates in the prefectures over 5-year periods ranging from 1916-1920 to 1971-1975. In addition, we calculated the ratio of the mortality rate of each prefecture to that of the entire country for each birth cohort.Results　Our APC analysis revealed that the decrease in the age-adjusted cancer mortality rates was mainly attributable to a reduction in the cohort effect on the rates in men and to reduction in the cohort and period effects on the rates in women. The magnitude of reduction in cohort effect varied by prefecture for men and women. Several prefectures having a government ordinance-designated municipality tended to show a higher reduction than those that do not. Spearman's correlation coefficient between the population size of prefectures and the percentage reduction in cohort effect was 0.370 in men. In addition, the relative ranking of the prefectures based on cancer mortality rates greatly varied by birth cohorts, particularly in men.Conclusion　A disparity exists in the percentage reduction in the cohort effect among prefectures. In each prefecture target cohorts with higher than average cancer mortality rates must be identified to implement specific countermeasures for cancer prevention. In addition, for each prefecture, assessment of lifestyle differences that might be related to cancer mortality among birth cohorts is important for reducing cancer mortality in the more recent birth cohorts.