| Abstract: | 在現行台灣全民健康保險的制度下，門診醫療至少有三大特色：平均就診次數高，就診於醫院（特別是醫學中心）的次數高，更換醫師與醫療院所頻繁。「逛醫院」有其社會文化與醫療制度的背景，應是目前台灣（或非西方先進國家）的特有現象。民眾非必要地常逛醫院，固然反映出其醫療需求尚未為現有制度所滿足，卻也易使得現有制度的運作效率大打折扣，亦會增加個人的機會成本與暴露於不必要的醫療風險，不僅有其總體健保財務意義，也是醫療制度管理改革的一大課題，值得研究。本研究計畫擬利用國家衛生研究院全民健康保險研究資料庫的檔案，運用資料探勘的技術，量化分析台灣地區民眾門診醫療的「逛醫院」現象。分析的面向將包括：病人人口學特性、醫療院所特性、醫師特性、疾病診斷、藥品品項、檢查檢驗項目等。近十年來，資料探勘主要運用於鉅量電子數據資料的分析上，其技術有別於傳統的統計分析，常見的技術包括frequent patterns, association rules, classification, cluster analysis, sequence analysis, social network analysis等，值得嘗試運用於全民健康保險研究資料庫龐大資料的分析上。台灣全民健康保險十餘年來的經驗，若能廣泛且縝密地分析，將有助於我國醫務管理研究學門的發展。深信以本研究計畫團隊過往的學術發表經歷，應能在此申請課題有深入且持續的國際論文發表。
Within the system of the National Health Insurance in Taiwan, the ambulatory delivery of health care exhibits at least three features: high utilization of physician consultations, high utilization of outpatient care (especially at academic medical centers), and frequent switching of consulted physicians and health care facilities. The phenomenon of 「doctor-shopping」 is closely related the socio-cultural environment and health care system. Although Taiwan is not the only country with such a phenomenon, the extent of 「doctor-shopping」 in Taiwan might be more serious than in other countries. The frequent visits to health care facilities by a person reflect the fact that her/his medical demand has not been met by the current system. On the other hand, such help-seeking behavior might lessen the efficiency of health care system, increase the personal opportunity cost, and have more exposure to unnecessary risk of medical care. The phenomenon of 「doctor-shopping」 not only causes a burden to the finance of health insurance, but also poses a challenge to the reform of health care management. In the current project, we plan to apply the techniques of data mining to quantitatively analyze the phenomenon of 「doctor-shopping」 in Taiwan, based on the nationwide claims data from the National Health Insurance Research Database of the National Health Research Institutes. The dimensions of analysis will include the patients, health care facilities, physicians, diagnoses, drug items, laboratory examinations, therapeutic procedures, etc. In the past ten years, data mining has been developed to analyze the gigantic amount of electronic datasets worldwide. Its techniques differ greatly from those of conventional statistics. The popular techniques of data mining include frequent patterns, association rules, classification, cluster analysis, sequence analysis, social network analysis, etc. The application of data mining to the data of the National Health Insurance might reveal more interesting and significant findings. The extensive analyses of Taiwan』s experiences in national health insurance system in the past decade will contribute to the development of health care management. The team members of the current project believe our records of academic publishing will guarantee in-dept studies about the phenomenon of 「doctor-shopping」 in Taiwan and continuous publishing at international journals. |
