Determining Minimal Clinically Important Difference of the Barthel Index in Stroke Patients

背景與目的：決定評量工具之最小臨床重要差異值(minimal clinically important difference, MCID)可幫助研究人員和臨床工作者判斷分數的改變是否達到臨床上重要的差異，以及協助解釋分數改變的意義。巴氏量表(Barthel Index, BI)是一廣泛運用於中風病人，評量其日常生活功能的量表，但目前BI量表的MCID值仍然未知，造成不易解釋BI量表分數改變的意義及應用。故此研究的目的是決定BI量表在中風病人的MCID值。 方法：此研究分成兩個部分：第一部份是追蹤研究，收集30位中風病人，以兩種病人自評方式分別估計BI量表的MCID值。第一種是在初評及再評時使用20公分之視覺類比計分表(20-cm visual analogue scale)讓病人自評日常生活功能狀態；第二種是在再評時使用15點李克氏量表(15-point Likert scale)讓病人自評日常生活功能改變情形。再由自評結果被歸為MCID組別的個案之BI量表分數改變量的平均值做為MCID的估計值，共產生兩個MCID估計值；第二部份是再測信度研究，收集56位慢性期中風病人，估計BI量表的再測信度係數值，以計算BI量表之測量標準誤(standard error of measurement)，並以測量標準誤做為第三個MCID估計值。前述三個估計值中，以最大的估計值做為BI量表之MCID值。 結果：在追蹤研究中，因為在視覺類比計分表和李克氏量表的自評結果被歸為具有MCID組別的個案，其在BI量表分數改變量差異大且樣本數少，故改採用分數改變量之中位數做為MCID估計值。30位病人中有8位在視覺類比計分表的自評結果歸為具有MCID的組別，產生第一個MCID估計值為2分。另外，30位病人中有14位在李克氏量表的自評結果歸為具有MCID的組別，產生第二個MCID估計值為1分。再測信度研究中估計出BI量表之測量標準誤，產生第三個MCID估計值為1.63分。因此，2分為BI量表在中風病人的MCID值。 結論：此BI量表的MCID值2分可幫助研究者解釋同一組中風病人在BI量表上的改變量以及不同組別間的差異量是否達到臨床上重要改變或差異，以協助臨床試驗或成效研究結果的解釋。而此MCID值2分亦可幫助臨床工作者判斷同一位中風病人在BI量表上的改變量以及不同病人之間的差異量是否達到最小臨床重要改變或差異，以協助相關臨床決策的制定或調整。Background and purpose: The minimal clinically important difference (MCID) of an instrument helps both researchers and clinicians determine whether the change scores achieve clinically important differences. However, the MCID of the Barthel Index (BI) was still unknown, which limited the application and interpretation of the change scores of the BI. Therefore, we aimed to determine the MCID of the BI in stroke patients. Methods: There were two parts of this study. First, 30 stroke inpatients participated in the follow-up study designed to determine the MCID of the BI with two anchor-based methods (using patients’ global ratings on the 20-cm visual analogue scale (VAS) and on the 15-pint Likert scale). The mean change scores on the BI of the MCID group based on the patients’ ratings on the VAS and the Likert scale, respectively, served as two estimates of the MICD. In the second part, 56 chronic stroke patients participated in the test-retest reliability study to determine the MCID of the BI with a distribution-based method. One standard error of measurement (SEM) served as the third estimate for the MCID. The largest MCID value of the three aforementioned estimates was suggested as the MCID of the BI. Results: In the follow-up study, the medians of BI change scores of the MCID group based on the VAS and the Likert scale served as two MCID estimates due to the large variation of the BI change scores and the small sample size of the MCID group. There were 8 patients in the MCID group that was based on scoring on the VAS, and the first MCID estimate was 2 points. There were 14 patients in the MCID group that was based on scoring on the Likert scale, and the second MCID estimate was 1 point. In the test-retest reliability study, the SEM was 1.63 points. Therefore, the MCID of the BI in stroke patients was found to be 2 points. Conclusion: The MCID of 2 points helps researchers determine whether the mean BI change/difference within/between stroke groups has reached the MCID, which is useful in explaining the results of the clinical trials or outcome studies. The value of 2 points also helps clinicians to determine whether the change/difference scores within/between individual stroke patients have reached clinically important change/difference and to make clinical decisions.Abstract...Ⅰ 中文摘要...Ⅲ Table of Contents...Ⅴ List of Tables...Ⅷ List of Figures...Ⅸ Introduction...1 Background...1 Introduction to the MCID...2 The importance of the MCID...3 The methods for determining the MCID...5 The anchor-based approaches...5 The distribution-based approaches...10 Other approach...14 Summary of the methods for determining the MCID...14 The importance of determining the MCID of the BI in stroke patients...15 The purpose of this study...16 Methods...17 Subjects...17 Procedures...18 Instruments...20 Data analysis...23 Results...25 Follow-up study (two anchor-based methods)...25 Sample description...25 Patients’global ratings on the 20-cm VAS to estimate the MCID...26 Patients’global ratings on the 15-point Likert scale to estimate the MCID...28 Test-retest reliability study (a distribution-based method)...29 Sample description...29 Using standard error of measurement (SEM) to estimate the MCID...30 Summary of the results...30 Discussion...31 The implications of the MCID of the BI...31 The strengths of this study...35 The limitations of this study...37 The methodological issues that remain to be addressed...40 Recommendations for further studies...44 Conclusions...45 References...46 Tables...60 Figures...66 Appendices...75 Appendix 1 Informed consent...75 Appendix 2 Barthel Index...76 Appendix 3 Mini-Mental State Examination...77 Appendix 4 20-cm VAS...78 Appendix 5 15-point Likert scale...8