中醫門診風險計價模式之建立

Abstract

爲控制醫療費用高漲並合理分配醫療資源，大部分先進國家的健康照護制度均已引入前瞻性預算支付制度，但爲兼顧健康保險的公平與效率性，採用論人計酬並配合風險校正機制的實施為有效方法之一。風險校正是利用個人的健康相關資訊作為風險校正因子來建立風險計價模式，藉此預測未來的醫療費用。近年來國內已有相當研究投入風險校正之探討，然而至今尚未有針對中醫所進行的風險校正研究，因此，本研究欲嘗試建立中醫醫療風險計價模式，並評估其對未來醫療費用的預測能力，以提供國內健保對於中醫部門前瞻性預算分配之參考。 本研究之資料來源為全民健康保險投保對象之基本資料以及2001年與2002年之申報資料，研究總樣本共計2,131,067人，運用人口因子、先前利用因子、診斷因子、診斷因子加上中醫利用偏好，共建構四個風險計價模式，風險計價模式之分析採weighted least square方式預測未來的醫療費用，其中診斷因子採用分類與迴歸樹（CART）法則建立分類組合，再進行後續預測能力分析。 研究結果顯示，人口因子模式預測力最低為1.24%，加入診斷資訊可使預測力增加至6.49%，再納入中醫利用偏好因素，模式預測力可大幅提升至17.51%，而先前利用模式之預測力最高為33.15%，在特定群體預測力方面，診斷偏好模式在投保地區分群的預測比除北區、東區及中區外，其他各區之費用預測具有一定之準確性，特定疾病分群預測比除少數疾病外，其餘疾病與1相差不大，模式大致上在疾病分群的預測比表現尚佳。 以診斷資訊及個人利用偏好所建立的中醫風險計價模式，可提升對未來醫療費用之預測力。由於中醫屬於我國固有的傳統醫療服務，與西醫間可能存在著輔助性或替代性，而民眾對醫療之偏好對其醫療服務之利用，實扮演極為重要之角色。然本研究僅單獨探討中醫之醫療利用，在預測費用上所反映的僅為民眾醫療利用的部分，故對於醫療預算設定所能提供之參考性，仍僅限於現況之不同部門的相加處理。因此，未來研究可同時考量中醫與西醫間醫療服務利用的相互關係，不僅可較完整瞭解民眾醫療資源利用之全貌，對於醫療預算設定更可提供整體性之參考。

In order to contain the escalation of health care expenditure and allocate resource fairly, many developed countries have introduced a prospective budget payment system. To maintain the equity and efficiency of health care, the risk-adjusted capitation approach is considered to be one of the effective mechanisms, in which health-based risk adjusters are used to predict personal medical expenses. Although abundant studies on risk adjustment have been conducted in Taiwan, none of them aims at the care of Chinese medicine. This study intends to develop risk assessment models for Chinese medicine and evaluate the predictability. The results can provide a reference of allocating Chinese medical budgets in Taiwan’s National Health Insurance. With 2001 and 2002 NHI data, 2,131,067 enrollees were selected as the study sample. Using demographic adjusters, prior utilization adjusters, diagnostic-based adjusters and a preference for Chinese medicine, this study develops four risk assessment models through weighted least square estimation and evaluates the predictability. Diagnostic-based adjusters are first classified by the Classification and Regression Trees（CART）. The results show that the risk assessment model based on solely demographic information has a poor predictability of 1.24%. Adding the diagnostic information into the demographic model can increase the predictability to 6.49%. Moreover, including the preference to Chinese medicine further increases the model’s predictability to 17.51%. Nevertheless, the prior utilization model has the highest predictability of 33.15%. The PR values of the diagnosis-preference model are close to 1, except for Northern, Eastern, Central districts and few diseases. The risk assessment model based on diagnostic information and the preference to the Chinese medicine can considerably improve the predictability. Because the Chinese medicine is a traditional medical practice in Taiwan, it may serve as a complementary or alternative medicine to the western medicine. The preference to different medical practices indeed plays an important role on individual’s medicine utilization. However, this study only analyzes the utilization of the Chinese medicine and provides information only relevant for setting additive budgets for different health care sectors, i.e., the current budget setting system. Further research on the relation with Chinese medicine and western medicine is encouraged and may provide the more complete panorama of utilization of medical care as well as budget setting.