運用健保資料庫分析全民健保第一類投保單位欠費特性

Abstract

全民健保之運作為隨收隨付制度(pay-as-you-go)，由全體保險對象、投保單位及政府共同繳納保險費至健保署，所有的醫療費用支出直接由大家繳納的保險費共同支付。我國醫療費用支出逐年上升，健保時常面臨收支不平衡及入不敷出之財務衝擊，而第一類投保單位的保費收入佔全民健保總保費收入約80%，大額欠費往往也是第一類投保單位所造成，故維持投保單位保費收入穩定一直是健保署收入面最重要的業務之一。本研究利用健保署倉儲資料庫，針對2015年1月到12月之間，累計欠費金額超過5,000元的第一類投保單位，並以隨機抽樣方式擷取未欠費投保單位後兩者合併，作為本研究所要分析的樣本資料，以摘要統計、決策樹及羅吉斯迴歸分析其影響投保單位欠費的特性，例如:行業別、單位規模、單位營運時間長短、負責人年齡層及受僱者投保金額等，可發現不同分析方式所預測的單位欠費結果不盡相同，最後透過誤分類率、ROC曲線以及累積增益圖作為評價不同分析方法之效益，找出可以預測第一類投保單位是否產生欠費之模型，讓主管機關能更了解欠費高風險單位的特性，做更有效率的投保單位保費繳納情形管控。

The National Health Insurance (NHI) program is operated based on the “pay-as-you-go” financing principle. The NHI system mainly derives its revenue from the premiums collectively paid by the insured people, employers, and the government, and the premiums paid for the medical expenditure. However, the revenue growth has lagged far behind the rate of increase of medical expenses, and the deficit of the NHI system has become a financial problem. About 80% of all premiums are paid by category 1 units, and they often caused large amount of overdue premiums as well. Maintanining a stable financial system of NHI is the most important responsibility of the NHI admistration (NHIA). Based on the NHI research database, this research collects the payment data and basic information of all category 1 units in 2015, and analyzes those data by using the decision tree analysis and logistic regression analysis. The analysis discovers the significant attributes of category 1 units which would have higher probability of not paying the premiums in time. The research findings can be used to bulid a more efficient premium monitoring system for NHI.