Data Mining for Seeking Relationships between Sickness Absence and Japanese Worker’s Profile

Since sickness absence causes great losses for both individual employees and their companies, decreasing sickness absence is one of major concern in occupational healthcare field. We conducted worksite-based study to elucidate the influence of lifestyle, medical findings and present illness on sickness absence of 1 week or longer during the fiscal year. Subjects were 6,010 Japanese male employees in a large telecommunication telephone company, aged 30 to 59 (the mean age 46.3±6.7 years), who had health check-up of in the fiscal year from 1991 to 1998 consecutively. We used data mining methods, such as ‘the Association Rule Analysis’, ‘the Correlation Coefficient Analysis’, and ‘the Risk Ratio Analysis’, to elucidate interrelationships in sickness absences, lifestyle (healthy / unhealthy), medical findings (normal / abnormal), and present illness (non-existent / existent), that were surveyed consecutively in the fiscal year from 1991 to 1998. In present illness, secular trend of Risk Ratio showed different patterns according to present illness category. Our results may contribute preventing sickness absence in Japanese worksite.