全民健康保險糖尿病醫療給付改善方案初步影響評估

The preliminary evaluation of Pay-for-Performance Program for Diabetes under National Health Insurance in Taiwan

李待弟(Tai-Ti Lee)

指導教授 ： 鄭守夏

國立臺灣大學/公共衛生學院/衛生政策與管理研究所/碩士(2007年)

https://doi.org/10.6342/NTU.2007.01970

摘要

目的：糖尿病是一種高發生率、高盛行率、高費用而容易導致併發症的長期慢性病，疾病帶來的經濟負擔，迫使健康政策需要採取積極的策略去減緩糖尿病對個人健康與生命的威脅，以及對社會的負擔。世界衛生組織預估至2025年全球將有三億三千萬名糖尿病患，2003年台灣地區糖尿病患人數已達780,000人（盛行率3.54%），且呈現逐年增加情形，醫療費用亦呈現逐年上漲。2001年底中央健康保險局實施「全民健康保險糖尿病醫療給付改善方案」，結合共同照護及論質計酬的慢性病照護模式為世界首創，實施效果值得分析，因此本研究主要目的是分析「全民健康保險糖尿病醫療給付改善方案」對糖尿病人醫療利用的初步影響，以實證分析結果回饋政策制定（support policy cycle）。

研究設計：本研究資料取自中央健康保險局1999-2002年全國性醫療申報檔，屬於次級資料分析，研究設計為準實驗研究設計，以糖尿病試辦方案實施的前一年（2001年）為前測期，實施的第一年（2002年）為後測期，並依本研究定義選取2002年參加試辦方案的病人為研究組，沒有加入方案的病人則為控制組，比較這二組在糖尿病試辦方案實施前後醫療利用的差異，以及分析影響醫療利用的因素。

研究對象：本研究對於研究對象的基本定義為，在一定期間，持續在台灣生活，具有常人持續就醫紀錄，以及糖尿病用藥較規則的一群人。因此研究個案必須為2000-2002年每年各有至少2次以上糖尿病診斷，及2001及2002年各有3個月以上糖尿病用藥，且2002年仍存活者。研究個案在2002年曾接受照護服務3次（含）以上者定義為研究組，共有樣本9,105人，完全沒有接受照護服務者共410,372人則屬於控制個案，再自410,372人中抽出20,726人作為分析樣本的控制組，以使二組樣本數相稱，合計分析樣本共29,831人。
結果：在醫療提供方面，以糖尿病相關疾病來看，加入糖尿病醫療給付改善方案的病人，在方案實施後7項重要檢查的平均完成種類提高，且達到統計顯著水準（P<0.0001）。而就7項檢查個別項目以差異中的差異統計法分析結果發現，在試辦計畫實施以後研究組病人接受各項檢查的平均次數增加均比控制組多。而在醫療利用方面，加入糖尿病醫療給付改善方案的病人在試辦計畫實施後一般門診次數增加（P<0.0001），但研究組與控制組之間差異不大；而在急診及住院的利用機會方面（probability of use），加入醫療給付改善方案的病人在試辦計畫實施以後因糖尿病相關疾病而住院與急診的機率比控制組明顯為低（OR=0.284，P<0.0001；OR=0.435，P<0.0001）。進一步再針對因糖尿病而住院的病人分析影響住院日數及住院件數的因素，結果發現在試辦計畫實施以後，研究組因糖尿病相關疾病而住院的日數及件數較控制組降低，且達到統計上顯著水準（P<0.0001）。醫療費用的部分，整體而言，試辦計畫實施以後糖尿病相關疾病平均每人總醫療費用是增加的（2,250元），增加的部分來自一般門診費用的貢獻，2001年與2002年研究組平均每人一般門診費用比控制組增加5,460元，其中包含一般門診就醫費用平均每人增加2,385元，以及管理照護費平均每人多出3,075元。但是相對於門診費用的增加，試辦計畫實施以後，因糖尿病相關疾病而急診及住院的費用則是減少的，分別為-20元及-3191元，因此進一步分析影響因糖尿病相關疾病住院費用的因素，結果顯示在試辦計畫實施以後，研究組因糖尿病相關疾病而住院的費用較控制組降低，且達到統計上顯著水準（P<0.0001）。另外，由糖尿病人「全部的醫療紀錄」分析結果顯示，2001年與2002年研究組與控制組，在7項檢驗檢查平均每人完成種類、7項檢查個別項目平均每人接受的檢查次數，門診、急診、住院平均每人利用量，及平均每人醫療費用的改變狀況，均呈現與「糖尿病相關疾病」的分析結果改變方向一致性的結果。
討論及結論：初步研究成果顯示，糖尿病試辦方案實施後7項重要檢查完成種類提高（compliance），住院與急診機率及利用則減少，顯示試辦方案對糖尿病人的照護品質有正向影響，但初步的成效只能代表在醫療給付改善方案大架構下達成提供較好的基本照護，需要更進一步的完整評估才能反映計劃所帶來對病人及整體醫療費用的影響。而細緻的醫療品質（quality）資訊健保局目前沒有完整、正確掌握，缺乏有效實證評估。目前台灣糖尿病論質計酬制度下沒有證據顯示為了得到好的照護成績而挑病人的情形。自2007年1月起糖尿病試辦方案新增血脂(Lipid)及血糖 (HbA1c)品質加成監測指標，建立獎勵評比制度，展現健保局改善的努力與進步。而本研究採用較嚴謹的定義選取樣本，可能因而犧牲一些較輕症或潛在病人，而缺乏可靠的檢驗品質資料，無法分析醫療結果（intermidiate outcome），另外醫療申報資料診斷碼的限制，致使就醫件數或費用無法百分之百釐清，亦可能影響糖尿病合併症或併發症之定義均為本研究之限制。由初步分析結果糖尿病醫療給付改善方案實施後對糖尿病人照護品有顯著的正向影響。

關鍵字

糖尿病 ； 品質 ； 醫療利用 ； 共同照護 ； 論質計酬

並列摘要

Objective : Diabetes is a chronic disease with high prevalence and high cost, which cause multiple complications, and incurs large financial burden. The government is force to adopt active tactics to reduse the threat to personal health and life of diabetes, and the burden to the society. World Health Organization estimated in 2025, there will be 330 million diabetes patients in the whole world. The number of diabetic patient in Taiwan has already reached 780,000 (prevailing rate is 3.54% ) in 2003, and their hospitalization cost also increase every year. The Bureau of the National Health Insurance implemented ' The Pay for Performance Program for Diabetes ' in 2001, combining ’shared care’ and ‘pay for performance’ mechanism. The program is novel to the world, its effect is worth analyzing. The main purpose of this study is to analyse the preliminary effect of ' The Pay for Performance Program for Diabete ' on medical utilization and expenses. Results of this study may provide feedback to the policy.

Design: Since the new payment program was implemented at the end of 2001, a before-after natural experimental design was used to assess the impact of the program. Diabetic patients who joined the new program were identified as the study group. Diabetic patients who never participated in this program were defined as the control group,only a portion of the patients were randomly selected for analysis. Observation periods were one year before (2001) and after(2002)the program was implemented.

Participants: The basic definition of the research subject is living and surviving in Taiwan, during the study period having active medical records and taking diabetic medicine. So patients in the study group must have diabetes diagnosis at least 2 times each year in 2000-2002, and have taken medication 3 months or more in 2001 and 2002, and who still survived in 2002. The study group in 2002 was composed of diabetic patients who have accepted the new services, there was 9,105 patients in all. And there were 410,372 patients defined as the control cases, who received no such services at all. A random sample of 20,726 subjects were selected to form the control group for analysis. The total number of patients was 29,831 in this study.

Results:(1) Patients in the study group tented to have higher proportion of completing the 7 important examinations for diabetes which reached a statistical significance level(P<0.0001). Using difference-in-difference analysis we also found that patients in the study group had more examination items executed for the 7 items than those in the control group. (2)In terms of the number of outpatient visits, the two groups showed no significant differences. However, the study group had lower probability of patient admission or emergence visits than the control group did. The inpatient length of stay and number of admission were also significanyly fewer in the study group than the control group (P<0.0001). (3)The healthcare expenditure tended to be mixed. Overall, diabtec patients spent more expenses (2250 NT Dollars) in 2002 than in 2001. Patients in the study group had higher expenses than the control group mainly due to the outpatient visit including the management fee provided by the new payment program. On the other hand, our regression model showed that patients in the study group had lower expenses for diabetes related inpatient services (3191 NT Dollars fewer, P<0.0001).

Discussion and conclusion: Our preliminary results showed that, after the implementation of the new payment program, the compliance rate for the 7 important examinations had increased and the probability of patient admission or emergency visits decreased. These results implied that the essential care to diabetic patients had been improved; however, more comprehensive evaluation was needed to detect the impact of the new program. Lacking detailed measures on quality of diabetic care we could not provide robust evidence on this issue. However, this study found no evidence showing that healthcare providers selected patients into the new program in order to receive more payment. The Bureau of NHI requested the healthcare facilities to provide new bio-medical data on Lipid and HbA1c which showed the strong ambition of the Bureau. Limitations of this study should also be mentioned: the rigid screening criteria for a diabetic patient used in this study might decrease the number of potential patients; lacking bio-medical data of the patients limited our analysis on treatment outcomes. The use of the NHI claim dataset without comprehensive diagnosis codes might affect accuracy in calculating expenses as well as the co-morbidity conditions. In summary, the preliminary findings indicated that the new pay-for-performance program had positive impact on the quality of diabetic care in Taiwan.

並列關鍵字

Diabetes ； quality ； utilization ； shared care ； pay for performance

參考文獻

行政院衛生署國民健康局.（2003）：糖尿病防治手冊 糖尿病預防、診斷與控制流程指引。台北：行政院衛生署國民健康局。

魏榮男,莊立民,林瑞雄,趙嘉玲,宋鴻樟.（2002）.1996~2000年台灣地區糖尿病盛行率與住院率. 台灣衛誌, 21(3),173-180.

賴美淑，邱淑媞（2002）.糖尿病共同照護之概念與內涵.台灣醫學,6(4), 560-567.

黃三桂，王悅萍，錢慶文（2002）.疾病管理對糖尿病患醫療資源耗用之影響。台灣醫務管理雜誌, 3:35-47.

王冠懿，鄭守夏等（2006）. 民眾就醫層級與醫療利用分析：SARS疫情前後的觀察.台灣衛誌,25(1)：75-82.