1、Evidence-Based Medicine 循证医学循证医学杨杨 学学 宁宁 Xuening Yang,MD广东省人民医院广东省人民医院 广东省肺癌研究所广东省肺癌研究所广东省循证医学科研中心广东省循证医学科研中心 循证医学杂志社循证医学杂志社Evidence-based Practice(EBP)Models Traditional practice Model()Model 1:Use the highest quality information to guide clinical decisions Model 2:Search,evaluate,and make availabl
2、e specialty specific best available evidence Model 3:Create original research or systematic reviewsEBM Practice-Five steps Patient History and PE Question Asking Answerable Clinical Questions.Search Searching for the best current evidence.Appraise Critically-appraising the evidence.Application:Integ
3、rating the evidence with our expertise and our patients unique biology and values Audit Evaluating our performanceStep 1 Asking Answerable Clinical Questions Setting:64 residents at 2 New Haven hospitals Method:Interviewed after 401 consultations Questions Asked 280 questions(2 per 3 patients)Pursue
4、d an answer for 80 questions(29%)Not pursued because Lack of time Forgot the question Sources of answers Textbooks(31%),articles(21%),consultants(17%)Green,Am J Med 2000Step 1 Asking Answerable Clinical Questions Most of our questions are never answered When answered,the information is likely to be
5、neither the best nor up-to-dateStep 1 Asking Answerable Clinical Questions EBM starts with a clear clinical question Well-built clinical questions guide the whole EBM process Type of question?Causation/etiology Intervention Diagnosis Prognosis Type of Study?What would be the best study design in ord
6、er to answer the question?study designs should limit biasStep 1 Asking Answerable Clinical Questions Question components:PICO What types of Participants?What types of Interventions?What types of Comparison?What types of Outcomes?Step 1 Asking Answerable Clinical Questions What types of participants?
7、Disease or condition of interest Potential co-morbidity Setting Demographic factors Task of clinical problemStep 1 Asking Answerable Clinical Questions What types of intervention?Types of treatment Type of diagnostic test Type of causative agent Type of prognostic factorStep 1 Asking Answerable Clin
8、ical Questions What types of outcomes?For causation,outcome is disease For diagnostic test,outcome in disease of interest For treatment include all outcomes important to people making decisions to define success of therapy For prognosis,outcome is the chosen endpoint of the diseaseStep 1 Asking Answ
9、erable Clinical Questions Broad or narrow questions Broad Do gefinitib improve survival in lung adenocarcinoma?Narrow Do gefinitib improve survival in femal lung adenocarcinoma refractory to first-line chemo?Step 1 Asking Answerable Clinical Questions CausationPeople-Exposure-Comparison-Outcomes Doe
10、s heavy smoker have higher incidence of lung cancer?Prognosis:People-Exposure-Comparison-Outcomes Is lung adenocarcinoma with high level of CEA are more likely to have recurrent disease?Step 2 Searching for the best current evidence Textbook Annually revised Use high level evidences With reference G
11、uidelines Cochrane Library MedlineNational Guideline Clearinghouse(http:/www.guideline.gov/)Step 2Too much?即使是小的问题也有相关指南Step 2Step 2Cochrane Library Advantage highest quality evidence well ever have on the effectiveness of health care Disadvantage not yet many systemic reviews necessarily omits the
12、newest treatmentStep 2EBP Models Traditional practice Model()Model 1:Use the highest quality information to guide clinical decisions Model 2:Search,evaluate,and make available specialty specific Level 1 information Model 3:Create original research or systematic reviewsStep 2PubMed/MEDLINE http:/www.
13、ncbi.nlm.nih.gov/PubmedStep 2Medline-PubMedAdvantage exhaustiveness,flexibility of searching journal coverage,currency(on-line versions)widespread availability and supportDisadvantage have to do your own quality filtering putting together good searches is difficult gaps in coverage(medical,geographi
14、cal and linguistic)Step 2Step 2Step 2Step 2Step 2Step 2Step 2 Searching for the best current evidence For primary literature Not just abstract!Try to get Full-textStep 3 Critically-appraising the evidence Valid?Design and methodology:logical?Sample Experiment or Measure:carried out in a sound way an
15、d Convincible？Follow-up rate?Analysis Conclusion Ethical?Importance?Clinical significance？Cost-effective？Applicable?Will the results help locally?Step 3 Critically-appraising the evidence Quick screening by title,abstract and relate message Authors Well-known？Past record？Creditable？Institute Compete
16、nce to do this study？Well-known？Specialty Has or had published in core journals?Your practice？Valuable？Go for Full-textDiscrimination？Rational？EBM？Step 3 Critically-appraising the evidence Type of QuestionSuggested best type of StudyTherapyRCTcohort case control case seriesDiagnosisProspective,blind
17、 comparison to a gold standardEtiology/HarmRCT cohort case control case seriesPrognosisCohort study case control case seriesPreventionRCTcohort study case control case seriesClinical ExamProspective,blind comparison to gold standardCostEconomic analysisRight methods is crucialThe Evidence PyramidHie
18、rarchy of evidence:arranges study designs by their susceptibility to biasPrimary Research Experimental observationalSecondaryResearchFor quantitative studiesFor qualitative studies No consensus about the relative rigour of different methodsSystematic reviewStep 3 Critically-appraising the evidencePh
19、ase III RCT head to head Compare survivalPhase III RCT head to head Compare survivalAdequate sample sizeStep 3 Critically-appraising the evidence Measure Objective：Lab test，Radiology Subjective：pain，symptoms Endpoint alternative：relieve of symptom,shrink of tumor Gold Standard：prolong survivalStep 3
20、 Critically-appraising the evidence Conclusion Based on the data?Based on the data?New?New?Updated?Updated?Compare with other study And,What next?Step 3 Critically-appraising the evidence Attention!Dont trust it just because it is from a big boss,meta-analysis,RCT,or guideline.Evidence should be in
21、consistence with known medical science(physiology,pathology,pharmacology,etc）Usage of scientific language is not necessarily mean scientific basedScientific methods creditNot all the evidences are the same importanceHollingworth&Jarvik,Radiology 2007;244:31-38Flowchart of cost-effectiveness at each
22、level of TA hierarchyTechnical PerformanceDiagnostic PerformanceDiagnostic ImpactTherapeutic ImpactImpact on HealthStep 4 Application Integrating the evidence with our expertise and our patients unique biology and values Your question Your patient Your expertise Your hospital How toStep 5 Evaluating
23、 our performance Follow-up Evaluate The patients benefit?Next Or again another 5 steps Collect data and write an articleEBP Models Traditional practice Model()Model 1:Use the highest quality information to guide clinical decisions Model 2:Search,evaluate,and make available specialty specific best av
24、ailable evidence Model 3:Create original research or systematic reviewsSelective EBP May not need to carry out all 5 steps 1.Asking Answerable Clinical Questions.X 2.Searching for the best current evidence.X 3.Critically-appraising the evidence.4.Application:Integrating the evidence with our experti
25、se and our patients unique biology and values 5.Evaluating our performanceSelective EBP 3 different modes of EBP Searching&appraising provides E-B care,but is expensive in time and resources Searching only much,quicker,and if carried out among E-B resources,can provide E-B care Replicating the pract
26、ice of experts quickest,but may not distinguish evidence-based from ego-based recommendationsSelective EBP Even fully EBM-trained clinicians work in all 3 modes Searching&appraising for the problems I encounter daily.Searching only for problems I encounter once a month.Replicating for problems I enc
27、ounter once a decade(and crossing my fingers!).Case1 Male,ZSU Professor，62 years old Small cell lung cancer,limited disease,chemo for 6 cycles with complete response Medical oncologist：prophylactic cranial irradiation（PCI）not imporve survival,and with AE！Radio-oncologist：PCI improve survival,and tol
28、erable Question:Why?Should the patient be give PCI?Improve survival?How much?Side effect？When?How?TextbookPublish 20061.Chance of SCLC brain met：50%-80%for 2 years2.1995，a META-analysis：SCLC with CR,PCI improve 3 year survival 15%-21%，but not analysis the complication3.Reduce brain met from 30%5%Con
29、clusion：SCLC with CR,PCI is recommend after 2 cycles of chemoComplications?case1Guideline&SR NCCN Lung cancer and Prophylactic cranial irradiation 1 2002 The Cochrane Library,Issue 2Prophylactic cranial irradiation Prophylactic cranial irradiation significantly improves survival and significantly im
30、proves survival and disease-free survival for patients disease-free survival for patients with small cell lung cancer in with small cell lung cancer in complete plete remission.case1例例1From Secondary Source PCI should start early after chemo complications：not RCT 2 study report no difference in ment
31、al or neurological disorder,but only study part of the study patients，40%and 60%（1995，1997）N=300 N=314 In one RCT（N=46）,report a mental disorder（1997）case1PUBMED Cranial IrradiationMESH AND lung neoplasmsmh(Cranial Irradiation/adverse effectsMESH OR Cranial Irradiation/mortalityMESH)AND lung neoplas
32、msmh25 papers，3 papers after 2006，on relate paper（level 5）例例1PUBMED 2009,JCO peer-review Complications Mental and neurological Increase by concurrent chemoirradiation Increase in dose level40Gy，no more than 30Gy，recommend 20Gy Support PCI in SCLC limited disease with chemo CR.例例1Conclusion&Decision
33、PCI reduce brain met,improve survival When RT 30GY,rare complications More benefit than harm Recommendation：PCI，right nowcase1Events Like This Sanlu baby formula milk powder contaminate with melamine Jul.16,2008 甘肃省卫生厅接电话报告，称今年该院收治的婴儿患肾结石病例明显增多，近几个月已达十几例，经了解均食用了三鹿牌配方奶粉。Sept.11,2008 新华社报道：“高度怀疑”三鹿公司生
34、产的三鹿牌婴幼儿配方奶粉受到三聚氰胺污染。case2Sanlu baby formula milk powder contaminate with melamine Sept.17,2008 国务院有关部门负责人和科学家 国家质检总局局长李长江：共检验了109家企业的491批次产品，有22家企业的69个批次的产品检出了三聚氰胺，检出不合格产品的企业约为20需要指出的是，这次被检出含有三聚氰胺产品的企业，不是所有批次的婴幼儿奶粉都有问题，比如，伊利、蒙牛、雅士利、施恩等企业被抽检的产品有几十个批次，但有问题的只有几个批次几个批次。case2Sanlu baby formula milk powd
35、er contaminate with melamine 卫生部陈竺部长：三聚氰胺主要对膀胱和肾脏有影响，引发动物膀胱炎、膀胱结石、肾脏炎症等，但但没有发现对人类有致癌作用没有发现对人类有致癌作用。婴幼儿食用出现不明原因的哭闹、呕吐、发热、尿液混浊、血尿、少尿或无尿等症状，应立即就近到医疗机构筛查就诊。食食用含有三聚氰胺的其他品牌婴幼儿奶粉用含有三聚氰胺的其他品牌婴幼儿奶粉，只有出现上述症状时，才需要立即到医院筛查就诊。从2008年9月12日至17日8时，各地报告临床诊断患儿一共有6244例。其中，4917例患儿症状轻微，生命体征稳定，正在进行院外随访治疗或已经治愈。现仍留院观察治疗患儿有13
36、27人。所有临床诊断病历中，有158人发生过急性肾功能衰竭，占总病例数的2.5%。经有效治疗，目前158例重症患儿当中，已经有94人病情平稳或已经治愈出院。此外，回顾性的调查发现有3例死亡病例，其中甘肃省有2名，浙江有1名，不包括在刚才提到的6244名诊断病例之中。case2The Evidence PyramidHierarchy of evidence:arranges study designs by their susceptibility to biasPrimary Research Experimental observationalSecondaryResearchFor qu
37、antitative studiesFor qualitative studies No consensus about the relative rigour of different methodsSystematic reviewSanlu baby formula milk powder contaminate with melamine Cell or animal is no equal to human But there are any corelation?No evidence does not mean no harm!How should we act?SummaryEBM is a Lifelong learning modellifelong,self-directed,problem-based learningKnow-how in finding,appraising,and applying evidenceFurther Reading on Evidence Based Medicine
