A Study of the Death Rate in Japan Based on Cause and Social Environment

Application of multivariate statistical analysis

Abstract

In order to ascertain the relationship between the cause of death and social environment, after a principal component analysis, the authors applied a varimax rotation to the statistical data obtained in 1970 and 1975, for all Japanese prefectures except Okinawa.
Twelve of the major causes of death, including lung and stomach cancers, and seven socio-geographical conditions were selected as variables. As a result of factor analysis, three factors were extracted.
The first factor (F1) accounted for about 30% of the total variance. It represented characteristic disease patterns, possibly due to urbanization. Prefectures could be classified into those where high death rates due to lung cancer accorded with the degree of urbanization, and into those rural areas where accidents were the major cause of death.
The second factor (F2) accounted for about 18% of the total variance. It represented disease patterns posibly based on temperature. Prefectures were classified into those where major causes of death included liver cirrhosis, tuberculosis and hypertension-i.e. western Japan-and those where stomach cancer and cardiovascular disease were the major causes-i.e. eastern Japan.
The third factor (F3) was characteristic of rural areas in northern Japan.
Standardized scores from (F1), (F2) and (F3) were plotted for 1975, and twelve prefecture types were obtained.
Death rates from the twelve diseases between 1972 and 1977 for the twenty-five cities in Yamaguchi Prefecture were calculated with reference to the five social environmental conditions measured by officical statistics for each municipality.