https://scholars.lib.ntu.edu.tw/handle/123456789/109498
標題: | 建立台灣診斷基礎風險計價模式及模式妥適性評估之國際比較 | 作者: | 張睿詒 | 關鍵字: | 健康保險;風險校正;診斷資料;DCG/HCC;ACGs;health insurance;risk adjustment;diagnostic information | 公開日期: | 31-七月-2005 | 出版社: | 臺北市:國立臺灣大學公共衛生學院醫療機構管理研究所 | 摘要: | 本研究係比較採用國外診斷群組分 類系統所建構之風險計價模式,及以本研 究所發展台灣本土診斷群組分類系統,所 建立台灣診斷基礎風險計價模式,運用在 台灣全民健康保險制度下,國內外不同診 斷群組分類系統所建立之風險計價模 式,對風險校正模式預測力之影響,及其 妥適性進行探討。 研究資料為國家衛生研究院全民健 康保險研究資料庫承保抽樣歸人20 萬人 資料檔,並選取承保資料中保險對象2000 與2002 年的全年納保之保險對象,研究 樣本為164,275 人,並擷取其門診及住院 醫療費用申報等相關資料。診斷群組建構 之風險計價模式,採用目前國際上廣泛被 採用之ACGs 模式(Adjusted Clinical Groups model) 、DCG/HCC 模式 (Diagnostic Cost Group/Hierarchical Condition Categories model),以及由本研 究所發展之門診診斷群組TASGs(Taiwan Ambulatory Spending Groups)與住院診斷 群組TPIPDCGs(Taiwan Principal Inpatient Diagnostic Groups),並預估各模式對2001 年及2002 年之醫療費用費用預測力;為 避免以相同資料進行建構模式與評估模 式預測力可能產生之過度估計問題,本研 究將研究樣本隨機分割為二子樣本,以估 計子樣本建立風險計價模式,而以預測子 樣本驗證所建立之風險計價模式的準確 性。 研究結果顯示,在門診費用預測力部 分, ACGs 模式為10.28%~11.2% 、 DCG/HCC 模式為8.85~9.49% 、 TASGs+TPIPDCG 模式為 20.37%~20.92%,在整體費用(門診與住院 合計) 預測力部分, ACGs 模式為 9.38%~10.56% 、DCG/HCC 模式為 7.44~8.29% 、TASGs+TPIPDCG 模式為 22.7%~24.76%,不論在門診費用或整體費 用之預測力,均以運用本土發展之診斷群 組分類系統預測力最高,而國外診斷群組 分類系統ACGs 模式預測力高於 DCG/HCC 模式。而ACGs 與DCG/HCC 模式運用在台灣全民健康保險制度下,其 預測力均相當穩定,顯示其運用於台灣全 民健康保險資料具相當的可適用性,惟國 外診斷群組分類系統均較本土發展系統 其預測力為低,顯示在導入國外診斷群組 分類系統時,應考量我國特性進行本土化 的修正。 This study intends to develop a diagnostic information risk adjustment model, and to examine the performance of ACG and DCG/HCC systems in Taiwan National Health Insurance program. Assess the extent to which ACG and DCG/HCC can account for same-year and next-year outpatient and total health care expenditures. A total of 164,275 beneficiaries, eligible for the National Health Insurance (NHI) from January 1, 2000 to December 31, 2002 were randomly selected from the enrollment file of National Health Insurance Research Database. ACG, DCG/HCC and TASGs+TPIPDCG were assigned through diagnoses from physician and hospital claim files. We use 2000 and 2001 diagnostic profiles to predict costs in 2001 and 2002, respectively. Linear regression was used to examine the predictability of the ACGs, DCG/HCC, TASGs+TPIPDCG systems to explain variation in individual costs. In order to avoid overfitting, a split-sample method was employed to partition the study sample into an estimation sample and a validation sample randomly. For Outpatient cost, while the ACGs system explained 12.87%~11.2%, the DCG/HCC system explained 10.24%~9.49%, and the TASGs+TPIPDCG system explained 20.92%~20.37% of variation in costs. For total cost, while the ACG system explained 15.43%~10.56%, the DCG/HCC system explained 11.5%~8.29%, and the TASGs+TPIPDCG system explained 24.76%~22.7% of variation in costs. The TASGs+TPIPDCG system performs better than ACGs and DCG/HCC system. The ACG system performs better than DCG/HCC system in Taiwan National Health Insurance program. The predictabilities of both ACG and DCG/HCC systems are fairly stable in the same-year and next-year costs. The results demonstrate that the application of ACG and DCG/HCC in Taiwan is feasible using existing data. Relative to the applications in other countries, the ability of the ACG and DCG/HCC systems to explain variation in costs is comparatively low. It is suggested that the introduction of the ACG and DCG/HCC systems into Taiwan requires some modifications in order to harmonize with local practice and disease patterns. |
URI: | http://ntur.lib.ntu.edu.tw//handle/246246/5033 | 其他識別: | 932416H002043 | Rights: | 國立臺灣大學公共衛生學院醫療機構管理研究所 |
顯示於: | 健康政策與管理研究所 |
檔案 | 描述 | 大小 | 格式 | |
---|---|---|---|---|
932416H002043.pdf | 56.03 kB | Adobe PDF | 檢視/開啟 |
在 IR 系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。