实症医疗照护决策.pptVIP

實證醫學之嚴格評讀 嘉義基督教醫院 整形外科 李孝貞 2010.11.18 醫學資料庫的種類 Secondary sources（證據評鑑資料庫） Cochrane database of systematic reviews Best Evidence, ACPJC, EBM Review, Bandolier (article review) Clinical Evidence (EB guideline or EB textbook) uptodate, MD consult Primary sources （原始文獻資料庫） 中華民國期刊論文, Medline, PubMed, CINAHL, …. 教科書, 網路電子書 The Evidence Pyramid Hierarchy of studies Read the “key” Levels of Recommendation (USPSTF) A – Highest – Strongly recommended (PAP smears) B – Recommended (Mammograms age 40+) C – no recommendation for or against (too close a balance between harm/benefit) (osteoporosis screening below age 60) D – Recommend AGAINST I – insufficient evidence to make any recommendation for or against (Prostate cancer screening) RCT (Randomized Control Trial) Lung cancer study 隨機取一群人，隨機分二組。一組每天至少抽煙一包共10年；一組不抽煙。若干年後統計兩組肺癌的發生率。 Cohort study 世代研究 Lung cancer study 找兩組人，一組有抽煙習慣，每天至少抽煙一包共10年，另一組不抽煙。若干年後統計兩組肺癌的發生率。 Case control study Lung cancer study 找兩組人 一組有lung cancer，分析其抽煙史(每天至少抽煙一包共10年)或不抽煙。 一組沒有lung cancer，分析其抽煙史(每天至少抽煙一包共10年)或不抽煙。 Case series Lung cancer study 找一組無 lung cancer無抽煙的人，每天至少抽煙一包共10年，若干年後得lung cancer的前後對照。 臨床問題的類別 Therapy select treatments that do more good than harm, worth the efforts and costs of using them Harm / Etiology identify causes of disease (including iatrogenic forms) Diagnosis (tests) select and interpret diagnostic tests Prognosis estimate the patients likely clinical course over time and anticipate likely complications of disease 依性質選最佳的研究方法 Different study design has different level of evidence, the best quality study design for each question: Frequency - Prevalence (case control), Incidence (cohort) Etiology / Harm - cohort Diagnosis – case control Prognosis - cohort Treatment / Intervention - RCT 實證醫學五步驟 步驟一： 形成出一個可回答的臨床問題 步驟二： 搜尋最佳證據 步驟三： 嚴格評讀證據之： (a)效度 (b)重要性（效益大小） (c) 臨床適用性 步驟四： 將臨床專業與病人價值觀相結合 步驟五： 評估執行效果及效用 步驟三：嚴格評讀證據: a)效度 b)重要性 c)臨床適用性 Validity (Reliability) 效度/信度 Can we believe it ? I