摘要:背景及目的:病人自陳醫療結果(patient-reported outcomes, PRO)可提供以個案為中心之主觀症狀、身體功能、安適感、及對醫療之滿意度。因此評量PRO有助於掌握中風病人之病情、協助設定治療計畫,進而提升臨床決策及療效驗證之效能。然而,當今評量中風病人之PRO至少有4項困難:(一)、多數PRO問卷所評量之內容(層面)不一,造成選擇評估工具困難。(二)、多數PRO工具之心理計量特性(如信度與效度)缺乏實證。(三)、多數PRO工具之最小重要差異值(minimal important difference, MID)及最小可偵測之差異值(minimal detectable change, MDC)未知,造成療效驗證研究之資料解釋困難。(四)、多數PRO工具之評估費時,造成個案及臨床人員之負擔。這些難題嚴重影響評量中風病人PRO 之有效性及可行性。羅序模式(Rasch model)可協助發展精準之評估工具,電腦適性測驗(computerized adaptive testing, CAT) 利用建構好之題庫,輔以電腦選擇最合適個案功能之題目測試。因此理論上,結合羅序模式及CAT 能夠快速且精準地評量PRO。因此我們5 年的計畫,將以羅序模式發展「中風病人自陳醫療結果電腦適性評量系統 (Computerized Adaptive Testing system for measuring Patient-Reported Outcomes of Stroke patients, CAT-PROS)以克服上述問題。我們並將證實CAT-PROS之心理計量特性及效能(如完成率與時間)。方法:第一年上半,我們依據過去研究結果及專家意見選擇5個PRO核心層面(主觀身體功能、能量/疲倦、主觀社會功能、情緒困擾及醫療滿意度),再依據相關學理、病人焦點團體及專家意見,以建構各層面100題,共500題。第1年下半年至第3年上半年,之後每題目訪問200病人,以利後續資料分析獲得穩定的數據。預計每位病人訪問100題(由500題隨機選出),故需訪問1,000位病人。第3年下半,我們將以結構方程模型(structural education model)及羅序模式檢驗資料之配適度(model-data fit)、單向度等特性。同時建立符合羅序模式及單向度特性之各項目參數,以成為CAT-PROS之題庫,並據以建立於個人電腦施測之CAT-PROS評量系統。我們再以模擬研究證實CAT-PROS之效能。第 4 年及第5 年,我們將驗證CAT-PROS 之心理計量特性及施測效率(時間及題數)。我們將追蹤評估CAT-PROS 及相對應之傳統評估工具於150位亞急性中風病人,以比較二者之心理計量特性(含信度、效度、反應性與MID)及效率,以確認 CAT-PROS優於傳統評估工具。另外,我們將以CAT-PROS評估200位慢性中風病人,並間隔2週執行再測,以驗證CAT-PROS之再測信度,並估計測量標準誤 (standard error of measurement)及MDC。同時我們將建立CAT-PROS之網路施測版本,並整合我們已發展的CATs(如評量平衡及日常生活功能),以提升CAT-PROS之方便性及實用價值。預期結果:因為採用最現代的測量理論與技術,我們預期CAT-PROS之心理計量將比傳統工具佳,且效能高,可於50分鐘或50題內施測完成,並可達到0.95以上之信度。本研究估算CATPROS各層面之MDC/MID值,將有助於臨床及療效驗證研究之資料解釋。價值及貢獻: CAT-PROS 將可提升臨床決策及療效驗證之效能。由於目前全世界並無類似評量系統,因此CAT-PROS 可能是國際上第一套。我們將結合已發展的CAT(如平衡功能及日常生活功能等),以整合成全面、完整的CAT評量系統。
Abstract: Background and purposes: Patient-reported outcomes (PROs) provide a comprehensive patient-centered approach to specifying symptoms, subjective functioning, well-being, and satisfaction with treatment in stroke patients. Thus, assessments of PROs are critical for patient management and outcome measurements.However, several challenges exist in measuring PROs for patients with stroke. First, most measures of PROs vary in content (domains), causing difficulty in the selection of relevant PRO measures. Second, most measures of PROs lack robust evidence of psychometric properties (e.g., reliability and responsiveness). Third, the minimal important difference (MID) and minimal detectable change (MDC) of most PRO measures for stroke patients are largely unknown, which limits the interpretation of change scores in clinical and research settings. Fourth, most PRO measures are time-consuming to administer, placing a burden of administration on both clinicians and patients. These fundamental issues greatly threaten the validity and feasibility of measuring PROs for patients, clinicians, and researchers.The Rasch model is useful in developing a precise measure. Furthermore, a computerized adaptive testing (CAT) system can provide efficient measurements. Therefore, in this 5-year project, we will develop a Computerized Adaptive Testing system for measuring Patient-Reported Outcomes of Stroke patients (CAT-PROS) to overcome the aforementioned challenges. We will determine whether the CAT-PROSis efficient and psychometrically robust. Method: We will first develop item banks for 5 core domains of PROs (i.e., subjective physical functioning, energy/fatigue, social function, emotional distress, and satisfaction with treatment) based on structural equation modeling and Rasch analysis for patients with stroke. We will select the 5 domains based on both core perspectives of health status and patients’ priorities as determined in our preliminarystudy. In the first year, we will construct 100 items for each domain based on focus group interviews of patients and clinicians, related theories and questionnaires, and suggestions from the experts. Then we will interview each item on 200 patients with various levels of severity until the third year of the project.In the third year, we will examine the model fitting of the items and calibrate the item characteristics with Rasch modeling. Then the CAT-PROS (a personal computer (PC) version) will be developed. We will perform a simulation study to determine the performance (e.g., efficiency and concurrent validity) of the CAT-PROS.In the fourth and fifth years, we will validate the CAT-PROS. The CAT-PROS will be administered on 150 sub-acute patients to validate its efficiency (i.e., feasibility and administration time) and psychometric properties (including reliability, validity, responsiveness, and MID). The CAT-PROS will also be administered on 200 chronic patients twice, two weeks apart, to determine its test-retest reliability, standard error of measurement, and MDC.Expected results: Since we use modern measurement theory (Rasch model) and techniques (CAT), we expect that the final CAT-PROS will reach an overall reliability above 0.95 (i.e., excellent reliability) and will require less than 50 minutes to administer or comprise less than 50 items (i.e., superb efficiency). The MID/MDC, which is critical for data interpretation, will be estimated for both clinicians and researchers.Potential value and contributions: The CAT-PROS will be useful for clinical decision-making, treatment planning, and outcome measurement for stroke patients. Because no CAT system measuring PROs exists for stroke patients, our system might be the first one to be developed. Furthermore, we will further combine our other developed CATs (e.g., balance, motor function, activities of daily living) in order to provide an integrative, comprehensive, and need-based assessment system for users.