국민건강보험 표본코호트2.0DB를 활용한 건강상태에 따른 암발생과 암수술건수 상대위험도 연구

A Study on Relative Risk of Cancer Occurrence and the Number of Cancer Surgeries Based on Health Conditions Using National Health Insurance Cohort 2.0DB

본 연구는 국민건강보험 표본코호트2.0DB을 사용하여 일반대상자와 간편고지대상자를 대상으로 남녀 각각에 대해 향후 1년 안에 예측된 암 발생률과 암수술건수를 적용하여 일반대상자와 간편고지대상자 간의 암발생 상대위험도와 암수술건수 상대위험도를 산출하였다. 일반대상자는 설명변수로 BMI, 혈압 등의 건강검진 항목들을 이용하고 간편고지대상자는 건강검진 항목과 직전 1년 장기투약 여부, 3가지 질병 치료 유무를 설명변수로 포함하였다. 암발생 예측은 로지스틱회귀모형을 이용하고, 암수술건수 예측은 음이항회귀모형을 이용하였다. 일반대상자과 간편고지대상자 간의 암발생 상대위험도와 암수술건수 상대위험도를 계산한 결과, 남녀 모두 암발생과 암수술건수 모두 일반대상자 대비 간편고지대상자의 위험도는 1보다 작게 나타났다. 이것은 가입조건을 완화한 간편고지만으로 유병자들을 포함한 보험가입대상자들의 상품영역 확대의 가능성을 보여주며, 또한 간편고지만으로도 암발생과 수술건수 담보의 상대위험도가 일반대상자 대비 확실히 구분됨을 실증적으로 제시하였다.

This study predicted cancer incidence rate and the number of cancer surgeries within the next one year for general subjects and simple notification subjects using the National Health Insurance Service Cohort2.0 DB. Relative risk of cancer occurrence and the number of cancer surgeries between simple notification subjects and general subjects were calculated. For the general subjects, health checkup items such as BMI and blood pressure were used as explanatory variables, and health checkup items, long-term medication for the previous year, and treatment for three diseases were included as explanatory variables for the simple notification subjects. A logistic regression model was used to predict cancer occurrence, and a negative binomial regression model was used to predict the number of cancer surgeries. As a result of calculating the relative risk of cancer occurrence and the number of cancer surgeries between the general subjects and the simple notification subjects, each relative risk of cancer occurrence and number of cancer surgeries in both men and women was less than 1 for the simple notification subjects compared to the general subjects . This shows the possibility of expanding the product range for those subject to insurance, including patients with pre-existing diseases, with only simplified notices that ease the subscription conditions. In addition, it empirically demonstrated that each relative risk of cancer occurrence and the number of cancer surgeries for simple notification subjects was clearly distinguished from that of general subjects.