請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/17686
標題: | 探討影響台灣成人糖尿病照護臨床治療慣性(Therapeutic inertia)之醫療提供者影響因素分析 Therapeutic inertia among adult DM patients in Taiwan focusing on associated factors of health care provider |
作者: | Li-Ying Huang 黃莉茵 |
指導教授: | 蘇喜老師 |
關鍵字: | 糖尿病論質計酬方案,糖化血色素檢驗值,加強給藥,臨床治療慣性,醫療提供者之主要變量,多層次分析法, Diabetes Pay-for-Performance Project,A1c,intensify therapy,therapeutic inertia,the variables of healthcare providers,Multi-level analysis, |
出版年 : | 2013 |
學位: | 博士 |
摘要: | 目的
針對慢性疾病糖尿病的治療,即使目前的實證或臨床指引指出積極的藥品治療可以減緩或預防併發症的發生,但部分醫師卻未給予適當的藥品治療,將疾病 (如血糖) 控制在理想的範圍內。此研究的目的首先為以糖尿病論質計酬方案 (Diabetes Pay-for-Performance Project, DM-P4P) 次級資料研究,根據病人糖化血色素檢驗值 (A1c),進行處方箋降血糖藥品改變的分析,並針對「加強給藥 (intensify therapy)」作為台灣「臨床治療慣性 (therapeutic inertia)」的型態研究 (pattern analysis)。其次,針對醫療提供者之主要變量,分析未給予「加強給藥」之重要影響因素。並以多層次分析方法,檢視不同醫療機構之層級別是否與「臨床治療慣性」相關。第三為根據「臨床治療慣性」之不同型態追蹤與A1c之治療結果的相關性分析。 研究設計 第一部分為診療前後是否介入處方改變,研究設計為介入前後比較研究;第二部份為「加強給藥」與糖尿病之主要結果指標A1c的關係,研究設計為針對參加2006年至2008年糖尿病論質計酬方案之糖尿病個案之世代研究 (cohort study)。 資料來源 此研究主要聯結兩個大型的資料庫,包括2006至2008年全民健康保險申報總檔 (the regular NHI claims database) 及糖尿病論質計酬方案之VPN (Virtual Private Network) 資料檔案。 統計分析 使用描述性分析及羅吉斯迴歸分析個案特質與醫療提供者之主要變量對於「加強給藥」的影響及「加強給藥」與糖尿病之主要結果指標A1c的關係。並運用多層次分析法 (multi-level analysis),探討在糖尿病照護中,檢驗不同醫院、醫師間的差異如何影響「加強給藥 (intensify treatment)」之因素。 結果 診療前後是否介入處方改變 在2006年至2008年加入糖尿病論質計酬方案,共215,679位糖尿病個案;1,527,539次的A1c數值,其中,僅25.98%的A1c數值低於7%的控制目標。針對A1c數值介於7%至11%,共有168,876位糖尿病個案,年齡為18至80歲者及899,135次的A1c數值。當該次A1c大於7%小於等於11%時,38.5% (346,221 處方箋數) 在診療後未獲得「加強給藥」。在藥品使用分類部分,42.0 %的藥品處方被歸為A1c值介入後新增加品項。在影響「加強給藥」之羅吉斯迴歸分析部分,年齡小於40歲之糖尿病個案處方箋較大於65歲之糖尿病個案處方箋數獲得「加強給藥」的發生比的1.26倍 (p< 0.001),糖尿病嚴重度 (DCSI score) 低的個案處方箋數較DCSI高的個案處方箋數獲得「加強給藥」的發生比的0.46倍 (p<0.001),糖尿病A1c為7%至8%的個案處方箋較A1c為10%至11% (10%<A1c=<11%) 的個案較不易獲得「加強給藥」;在醫師層次特質中,不同的醫師的專科別會影響到是否會給予病人「加強給藥」;在醫療機構特質中,醫院之層級別醫學中心較基層診所在給予「加強給藥」發生比為1.175倍 (p<0.001)。運用階層線性模式分析對於個人層次、醫師層次與醫院層次的特質是否會影響糖尿病個案是否會得到「加強給藥」部分,茲將研究結果與討論敘述如下。在個人層次部份,研究結果顯示: 首先,共有899,024次糖尿病個案處方箋數資料,巢套於20,729位醫師及巢套於3,422間醫療機構。 (一) 糖尿病個案的特質會影響是否會得到「加強給藥」部分 個人層次變量對接受到「加強給藥」的解釋量R2為1.18%。男性較女性不易接受到「加強給藥」,當控制其他變量時,DCSI score為0分者較DCSI score為4分者在接受「加強給藥」的機會少56%;當個案的年齡少於65歲較年齡大於65歲者易接受到「加強給藥」;A1c小於等於8%者,接受到「加強給藥」的機會較A1c大於11%者少70%。 (二) 醫師層次特質中會影響是否會給予「加強給藥」部分 醫師層次變量對接受到「加強給藥」的解釋量R2為30.78%。醫師年資少於10年的醫師在給予個案「加強給藥」的機會較年資大於30年的醫師多158%,年資少於10年較年資大於30年較易給予個案「加強給藥」。在醫師專科別部份,心臟內科醫師較新陳代謝科醫師較不易給予個案「加強給藥」,但未達到統計上顯著的差異。 (三) 醫院層次特性會影響是否會給予「加強給藥」部分 醫院層次變量對接受到「加強給藥」的解釋量R2為18.97%。台北健保分局較南部健保分局不易給予個案「加強給藥」。當控制其他變量時,醫學中心較基層診所較易給予個案「加強給藥」多48.20%。 「加強給藥」與糖尿病之主要結果指標A1c的關係 共有41,948位病人,皆觀察6至18個月,且至少有兩次的A1c值。在糖尿病用藥部份,起始A1c時接受「加強給藥」者,占了28,306人 (67.5%)。且接受「加強給藥」較未給予「加強給藥」,經6至18個月觀察後,會使得「A1c治療結果大於9%」發生比為0.779倍,發生比小於1,具統計上顯著意義,(p< 0.001)。在A1c檢查次數部分,在6至18個月內,若病人的A1c檢查次數每增加一次,「A1c治療結果大於9%」的發生比為0.972倍(p<0.001)。接受「加強給藥」的病人,相較於未接受「加強給藥」的糖尿病個案,經6至18個月觀察後,照護結果A1c 數值改善0.131% (p< 0.001)。 結論 臨床慣性有多種來源,而且會受到病人與醫療提供者等主要相關因子影響。在病人因子包含:年齡、DCSI score、A1c。在醫療提供者等主要因子包括:醫師年資、專科別、醫院之層級別及健保分局別等。本研究提出一項可用以量測此現象的品質計量法。有趣的是,高的血糖數值的確是「加強給藥」一項重要的預測因子。另外,在此研究發現接受「加強給藥」的病人比較容易在血糖控制上達成較佳的預後結果。在品質的評估中,建立此等「過程」與「預後指標」間的關聯是一項相當重要的目標。這樣的關聯進一步證實此研究中的「加強給藥」計量法的效度,也暗示在此研究中已經捕捉到一個重要的照護面向。這些結果也凸顯欲改善醫療照護的介入方法應著重於臨床慣性此一問題。 Background Empirical evidence and clinical guidelines related to the treatment of diabetes have indicated that aggressive drug therapy can mitigate or prevent the occurrence of complications; however, many physicians fail to prescribe appropriate drug therapy for control of the disease (such as blood glucose levels). This study is divided into three stages. Initially, we present analysis of secondary data related to the Diabetes Pay-for-Performance Project (DM-P4P). We used hemoglobin A1c (A1c) test values to identify variations in the prescription of hypoglycemic drugs. We then conducted pattern analysis on intensive therapy to elucidate the issue of therapeutic inertia in Taiwan and examined the variables of healthcare providers to analyze the factors associated with intensive therapy. Multi-level analysis was performed to determine whether the level of the medical institutions was correlated with therapeutic inertia. In the third stage, we conducted correlation analysis on the patterns of therapeutic inertia and the outcome of A1c. Research Design The first stage involved identifying changes in prescription patterns and conducting a comparison of prescriptions before and after A1c intervention. In the second stage, we determined the relationship between intensive therapy and the primary index of A1c. This involved a cohort study on diabetes patients that participated in DM-P4P from 2006 to 2008. Data Source We accessed two major databases: the regular NHI claims database from 2006 to 2008 and the virtual private network (VPN) of DM-P4P. Statistical Analysis This study employed descriptive analysis and logistic regression to determine the influence of variables related to primary healthcare provider on intensive therapy and the relationship between intensive therapy and the primary index of diabetes, A1c. Multi-level analysis was also conducted to examine whether different hospitals or physicians influenced intensive diabetes therapy. Results A total of 215,679 diabetes patients participated in DM-P4P between 2006 and 2008. Among the 1,527,539 A1c test results, only 25.98 % were less than 7%. We focused on the 168,876 diabetes patients presenting A1c values between 7% and 11%. The age of the study population ranged between 18 and 80, and 899,135 A1c values were included. Among patients with A1c values greater than 7% and equal to or less than 11%, 38.5% (346,221 visits) did not receive intensive therapy. Of the total drug prescriptions, 42.0 % were categorized as newly added drugs following A1c value intervention. In multivariate analysis of the factors related to intensive therapy, the odds ratio of a diabetes patient under 40 years of age receiving intensive therapy was 1.26 times that of a diabetes patient over 65 years of age (p< 0.001), and the odds ratio of patients scoring zero on the Diabetes Complications Severity Index (DCSI) receiving intensive therapy was 0.46 times that of patients scoring 4 on the DCSI (p< 0.001). Diabetes patients with A1c of 7% to 8% were less likely to receive intensive therapy than patients with A1c values of 10% to 11% (10%<A1c=<11%). The specialization of physicians also influenced whether the patient was administered intensive therapy. The odds ratio of medical centers providing intensive therapy was 1.175 times that of primary clinics (p< 0.001). This study performed hierarchical linear analysis to determine whether characteristics at the individual level, the physician level, and the hospital institution level influenced the likelihood of diabetes patients undergoing intensive therapy. Our results and a brief discussion are as follows. In terms of the individual level, the study results indicate the following. First, we obtained data from a total of 899,024 patient visits nested within 20,729 physicians nested within 3,422 medical institutes. 1. Influence of patient characteristics Variability at the individual level presented explanatory power of R2 = 1.18 % for intensive therapy. Male diabetes patients were less likely to receive intensive therapy than female patients, and when other variables were controlled, patients with a DCSI score of 0 had a 56 % less chance of receiving intensive therapy than those with DCSI scores of 4. Patients under 65 years of age were more likely to receive intensive therapy than those over 65 years of age, and patients with A1c results less than or equal to 8 % were 70 % less likely to receive intensive therapy than those with A1c results greater than 11%. 2. Influence of physician level characteristics Variability at the physician level presented explanatory power of R2 = 30.78 % for the intensive therapy provided. The odds ratio of physicians with less than 10 years of experience administering intensive therapy to patients was 158 % higher than that of those with over 30 years of experience; therefore, physicians with less than 10 years of experience were more likely to provide patients with intensive therapy than those with over 30 years of experience were. In terms of the specialty of the physician, cardiologists were less likely to administer intensive therapy to patients than endocrinologists were; however, this difference did not reach statistical significance. 3. Influence of medical institution level characteristics Variability at the medical institution level presented explanatory power of R2 = 18.97 % for intensive therapy that was administered. The northern division of the Bureau of National Health Insurance (BNHI) of Taiwan was less likely than southern division to administer intensive therapy. When the other variables were controlled, medical centers were 48.20 % more likely than primary clinics to provide intensive therapy. Regarding the effects of intensive therapy of hyperglycemia on A1c outcome in DM, a total of 41,948 patients were observed for 6 to 18 months and underwent at least two A1c tests. In terms of diabetes medication, 28,306 patients already undergoing intensive therapy after the index A1c test accounted for (67.5 %). After 6 to 18 months of observation, the odds ratio of a patient receiving intensive therapy obtaining an A1c result greater than 9% was 0.779 times that of a patient not receiving intensive therapy. This figure is less than 1 and thus statistically meaningful (p< 0.001). In the number of A1c tests taken, the odds ratio of a patient receiving an A1c result greater than 9% was 0.972 times for each additional A1c test taken (p<0.001). After 6 to 18 months of observation, the A1c results of patients undergoing intensive therapy were 0.131 % better than those of patients not undergoing intensive therapy (p< 0.001). Conclusion Therapeutic inertia for diabetes may arise from a number of sources; however, the primary factors are associated with the patient and the healthcare provider. Patient-related factors include age, the severity of diabetes (DCSI), and A1c value; factors related to the healthcare provider include the experience of the physician, their specialty, the level of the medical institution, and the BNHI division. This study proposed a practical approach to the measurement of quality in order to quantify therapeutic inertia. What is interesting is that high blood glucose levels are a crucial predictive factor of intensive therapy and patients receiving intensive therapy were more likely to achieve favorable prognostic outcome in the control of blood glucose. In the evaluation of quality, establishing a connection between the process and the prognostic indicator is crucial. Such a connection verifies the validity of this intensive therapy measurement and also implies that this study has captured a vital aspect of healthcare. These results also emphasize the importance of considering the issue of therapeutic inertia when implementing intensive therapy as a means to improve medical treatment. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/17686 |
全文授權: | 未授權 |
顯示於系所單位: | 健康政策與管理研究所 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-102-1.pdf 目前未授權公開取用 | 6.6 MB | Adobe PDF |
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。