全民健保自民國84年開辦後,至今已累積近10年的健保資料。如何透過已累積經驗,從中學習控制逐年高漲的健保費用是衛生當局的一個重要課題。本文從一個健保資料研究者的角度,討論健保資料庫的優點和缺點,使用原則和操作技巧,以及潛力和限制。作者首先透過和其他現有醫療資料庫的比較,使讀者瞭解健保資料庫的優缺點。由於健保資料庫檔案衆多,依據各檔案特性,作者先將健保資料庫區分為三大子資料庫:費用檔,醫令檔,和基本資料檔;其次,針對各子資料庫的整理、各子資料庫間的串結、以及健保和非健保資料連結這三部分,說明健保資料的處理原則。最後,透過一個研究實例,詳細解釋如何將處理原則和實務操作相結合,並提供一些研究方向供參考。
Since the inception of National Health Insurance in 1995, Taiwan has accumulated a decade of National Health Insurance Data (NHID). This data can assist the health authority to learn to help control the rising health spending. From a researcher's perspective, this paper discusses advantages and disadvantages, possibilities and restrictions, as well as principles and skills for using NHID. I first show the strength and weakness of NHID by comparing NHID with other important health data sets in Taiwan. Because NHID consists of many data files, I regroup these files into three sub-databases based on each file's specific role: expenditures, medical orders, and basic characteristics of patients, providers and physicians. Then I describe the rules concerning how to analyze each sub-database, how to merge files from two different sub-databases, and how to connect NHID with other non-health data. Finally, I use a research example to explain in detail the process of data construction, and discuss directions of future research.