| 摘要: | 由於醫療照護費用之高漲，讓醫管人員對於控制醫療利用之興趣增加。消化性潰瘍出血並非一罕見疾病，而治療消化性潰瘍出血之成本更是一巨大花費。據估算每年約7億元約佔健保支出0.24%。雖然有一大堆被接受之臨床指引來引導臨床醫師如何治療這些病人，但對於病人治療過程當中，仍有在用藥與治療處置上相當多之選擇差異性，因而在不同之背景與不同醫師之間，對於治療消化性潰瘍出血病人之治療處置上即有相當大之差別，並意味著在這治療處置之差異性上，有降低醫療成本的空間。一般說來，降低一特定性疾病醫療照護成本意味著須著手介入高醫療利用者之分析。因而了解這類病人是什麼特性導致在治療處置上選擇之差異性是必要的。 本研究目的是利用CART在控制疾病嚴重度後，檢定全國消化性潰瘍出血病人住院就醫時，醫院之層級別、權屬別、診治專科醫師之年資與科別、病患年紀、性別及其疾病嚴重度（以病人接受輸血之血液血漿費用和合併內科其他疾病來作測量代表）等自變項對依變項之總醫療平均費用之影響，並瞭解出血病人醫療資源耗用分佈情形。 我們的研究對象是民國89年因消化性潰瘍出血而住院之全國病患，並登陸於國家衛生研究院發行之全民健康保險研究資料庫中，在主診斷欄位其ICD-9編號為 5310, 53100, 5320, 53200, 5314, 53140, 5324, 53240, 53300任一者，在刪除資料不齊全者與極值後，總共有13,581個案進入本院研究，並透過單變項分析、ANOVA test統計分析與分類迴歸樹狀法為分類工具。 將在89年曾因消化性潰瘍出血住院病患其所有醫療總費用平均為預測準則，以醫師其個人工作機構之特徵如醫院之層級別、權屬別，醫師個人之特性如醫療專科別、專科醫師年資，及病人之特性如年齡、性別、接受輸血之血液血漿費、其他相伴隨之內科合併疾病數目、就醫科別做為分類準則，以CART進行所有可能之影響變項，作二元樹狀之分類。在所有自變項中以疾病嚴重度（以病患之輸血血液血漿費用和合併內科其它疾病來作測量代表）乃評估總醫療費用最有意義之變項，而病患就醫時科別、醫院層級別與病人年齡也是重要之影響變項。在經CART分為15個同質性醫療費用相似之子群，而綜觀總醫療費用，可發現費用最高可達306,478元，最低為15,495元。其中，病人若接受超過15,250元之血液血漿費時其總醫療費用最高。相反地，病人若接受小於1,525元之血液血漿費，小於4個合併內科其它疾病，就醫科別為腸胃內科、內科或家醫科時其總醫療費用最低，僅有15,495元。意味病患之疾病嚴重度在醫療費用扮演非常重要角色，而專科醫師因為個人執業之偏好與專家背景之不同，對病患處理之方式也不同，對醫療費用也有影響。 本研究成功地應用新方法-CART於醫學研究上，藉助迴歸分析法將資料分類並以決策樹方式呈現分類結果，並可去除自變項彼此間之相互作用。而不同醫師在醫療費用上有意義差異，意味醫療照護品質與醫療資源之利用須被監測，以期降低不必要之醫療處置與並合理利用醫療資源。因此醫管人員如何能加強臨床醫師之規劃訓練與經驗傳承，並灌輸醫師對醫療成本和品質之觀念，使其醫療行為上能有所規範，將是對醫療資源的使用合理化提供良好之基石。至於何種醫療費用細項導致彼此差異，值得將來我們更進一步之探討。; The growth of health care expenses increases interest in controlling medical utilization. Bleeding peptic ulcer (BPU) is not a rare disease in Taiwan. In addition, the costs of treating BPU are substantial. Although a widely accepted guideline direct physicians how to treat those patients, there still are several selections in medicines and procedures. As a result, the procedures selected to treat the patients of BPU are quite different from setting to setting and from physician to physician. The variation of selection of treatment procedures for patients indicates that there is a room for reducing health care cost. Generally, interventions for reducing health care cost for a specific disease are implemented for high-utilized group. Therefore, we must realize that what characteristics result the variation of selection of treatment procedures. Current study tests the impacts of independent variables such as the levels of hospital accreditation, ownerships, the years of practice, type of specialist of physicians, age, gender, and severity of illness (blood/plasma fee and accompany medical diseases) on the dependent variable, grand total medical expenses (GTME). Our data was collected from the database of National Health Insurance in Taiwan and total 13,581 inpatients of BPU in 2000 were included. Our inclusion criteria were cases with the major diagnosed as ICD-9 5310, 53100, 5320, 53200, 5314, 53140, 5324, 53240, or 53300 and excluded those cases with missing data or extreme value. The exploratory methods, classification and regression tree (CART), univariate and ANOVA test were utilized as analytic techniques. Our result show that severity of illness, measured by blood and plasma fee and accompany medical disease, was the most significant variable to predicate the grand total medical expenses. In addition, the levels of accreditation, type of specialist and age of patient are also significant variable. Fifteen homogeneous groups with similar grand total medical expenses were obtained from the analysis of CART. Total grand medical expenses range from as high as 306,487 NT dollars to as low as 15,495 NT dollars. Patients, with more than 15,250 NT dollars of the fee of blood and plasma consumed the most expenses. On the other hand, patients, with less than 1,525 NT dollars of the fee of blood and plasma and less than 4 accompany medical diseases, treated by internal medicine and gastroenterologist or family physician consumed only 15,495 NT dollars. We have presented a new approach to detect the effects of independ |
