The influence of lifestyle-related disease on colon cancer by the decision tree analysis using DPC data

Abstract

We investigated the relations among age, gender, smoking information, and lifestyle-related disease （hypertension :HT, type 2 diabetes ;DM, hyperlipidemia : HL, body mass index: BMI）in cases with colon cancer(disease cord:060035&060040), using Diagnosis Procedure Combination survey data（DPC data）. The risk factors to affect colon cancer was investigated using a method of decision tree. As a result, HL was extracted as the most significant factor relating to colon cancer. The rate of colon cancer patients without HLwas 2 times higher than the rate of with HL . The second factors was HT. The rate of colon cancer patients with HT was 3 times higher than the rate of without HT. After HL was excluded from an independent variable, the decision was finished. Moreover, colon cancer was divided into 060035 and 060040, then analyzed them under the same conditions. As a result, 060035 was the most strongly related to gender. Furthermore, both men and women were relevant to age. After gender was removed from the items of independent variable, 060035 had relevance to BI and then related to BMI. On the other hand, 060040 was only related to HL. HL was the most strongly related to colon cancer and 060040. However, the related items of 060035 were completely different. For this reason, ｗe consider that it is necessary to analyze colon cancer for each site.