腦中風商業保單設計與評價：全民健保資料之運用

Designing and Pricing Commercial Stroke Insurance: an Application of Taiwan NHIRD

由於近年日益升高的腦中風危害，本研究依據全民健保資料庫，估計國人腦中風發生機率與相關醫療成本，設計腦中風保單，並進行保費精算。研究追蹤全民健保資料庫自1996年至2012年，共99,165位40歲以上投保人之資料，分別以追蹤羅吉斯迴歸模型 (panel logit model) 推估腦中風機率，與追蹤負二項迴歸模型 (panel negative binomial model) 推估因腦中風就醫/住院次數。研究發現，年齡、性別、月薪資、居住地區、職業別與疾病史，對腦中風機率和就醫/住院次數有顯著的影響。其中年齡、男性、糖尿病、高血壓、牙周病、腎病、心房顫動、弱勢族群、地區人口對於中風機率影響皆為正向；而月薪資較高、職業為軍公教或農漁民者，影響為負向。居住於南部、東部或離島地區者，中風機率均較北部居民低。邊際效果來看，年齡每增加1歲，中風機率平均增加0.14%，男性中風機率平均較女性高出0.60%。此外，罹患糖尿病、高血壓、腎病、心房顫動的患者，中風機率皆增加超過1%。最後，本文為台灣保險公司與民眾設計一張腦中風商業保單，根據此實證模型，預估腦中風機率和次數進行保費估算，藉此提高國人因腦中風而遭受危害的財務保障。

 

Given the increasing incidence of stroke, we design and price a commercial stroke insurance using National Health Insurance Research Database (NHIRD) for insurance companies and customers in Taiwan. We collect 99,165 people over 40 years old during the period from 1996 to 2012 to perform the empirical study. In the study, the panel logit model and panel negative binomial model are employed to estimate the probability of stroke and the frequency of outpatient visits or inpatient days due to stroke respectively. We find that age, gender, monthly income, place of residence, occupation, and disease history have significant effects on the probability of stroke and the frequency of outpatient visits or inpatient days due to stroke.