服務量對醫療結果的影響—以台灣的肝癌病患為例 ; The effect of volume on treatment outcome: evidence from liver cancer treatment in Taiwan

NCUIR > School of Management at National Central University > Graduate Institute of Industrial Economics > Electronic Thesis & Dissertation > Item 987654321/44172

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/44172

Title:	服務量對醫療結果的影響—以台灣的肝癌病患為例;The effect of volume on treatment outcome: evidence from liver cancer treatment in Taiwan
Authors:	江尚峰;Shang-Feng Chiang
Contributors:	產業經濟研究所
Keywords:	工具變數;內生性;醫療結果;服務量;volume;outcome;endogeneity;instrumental variables
Date:	2010-07-05
Issue Date:	2010-12-08 14:52:51 (UTC+8)
Publisher:	國立中央大學
Abstract:	本篇論文旨在探討醫院、醫師服務量對於醫療結果的影響。本文利用1996年至2007年全民健保資料庫之「2000年承保抽樣歸人檔」進行研究，並挑選初次接受住院治療的肝癌病患作為研究對象。我們的醫療結果是以病患是否於一個月內、六個月內及一年內發生死亡作為定義，且分別建構醫院、醫師的前期服務量及當期服務量，以衡量經驗累積、規模經濟對醫療結果的影響。此外，鑒於品質較佳的醫院或醫師會吸引較多的病患前來就診，致使服務量產生自我選擇的內生性問題，因此本研究將藉由IV（工具變數）和2SLS（兩階段最小平方法）來矯正內生性所造成的偏誤和不一致。本研究的實證結果發現，在未考量到內生性的情況下，醫院、醫師的當期服務量與一個月、六個月及一年死亡率之間，大多呈現顯著負相關，此表示醫院和醫師的醫療行為具有規模經濟。但是當我們把前期服務量放入迴歸式之後，雖然醫院、醫師的當期服務量仍具有顯著性，但其前期服務量與醫療結果之間皆無明確的關係，是故醫院及醫師的醫療行為不具有學習效果。在進行2SLS之前，我們會對工具變數進行弱工具變數檢定及過度確認檢定，而檢定結果也顯示本文所選取的工具變數不為弱工具變數，且符合除外限制，故滿足工具變數於使用上的假設。最後，在考量到內生性的估計結果之中，我們僅發現醫師的當期服務量越高，其醫療結果越佳，反觀醫院當期服務量對醫療結果並無顯著影響。This paper uses longitudinal data on hospitalized patients who received liver cancer treatment to examine whether hospital or physician volume affect treatment outcome in Taiwan. Data set was obtained from National Health Research Institute, which is a twelve-year claimed data (1996-2007) of National Health Insurance. Our measure of treatment outcome is the probability of death in 1, 6 and 12 months after the discharge from hospital. We construct a set of current and one-period lag volume provided by hospital and physician to capture scale economies and learning effect on the treatment outcome of liver cancer. Because hospital and physician volume are endogeneous, i.e. hospitals or physicians who provide better quality care will attract more patients. We use instrumental variables (IV) and two-stage least squares regression to solve the bias from endogeneity. The results show that mortality decreased as both current hospital and physician volume increased after the endogeneity has not been controlled. But there is no evidence that learning effect (one-period lag volume) influenced the treatment outcomes. When we use IV to control for endogeneity, current physician volume was inversely related to mortality for the treatment outcome of liver cancer.
Appears in Collections:	[Graduate Institute of Industrial Economics] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	1748	View/Open

社群 sharing

Loading...