Corso di introduzione all'evidence Based Medicine2

asmn.re.it

Corso di introduzione all'evidence Based Medicine2

PROGRAMMA DELLA GIORNATA• si riprendono i concetti della volta precedente• “compiti” dati da fare a casa• si richiama il programma del corso• 1.1.1. Studi integrativi1.1.2. Revisioni sistematiche• 1.1.4.1. TEST DIAGNOSTICI• 1.1.5.CochraneBanche dati di revisioni sistematiche• Lettura di una revisione sistematicabaricchir Dipartimento di PatologiaClinica ASMN


Systematic reviewsand meta-analysisbaricchir Dipartimento di PatologiaClinica ASMN


The need for reviews• The explosion of biomedical publishing• Many of the studies give unclear orcontradictory results• Each trial may offer little information , thehope is that , when taken together a clearerpicture will emergebaricchir Dipartimento di PatologiaClinica ASMN


When systematic reviews are needed ?• Systematic reviews are needed wheneverthere is a substantive therapeutic question ,several primary studies -perhaps withdisparate findings – and substantialuncertainty.baricchir Dipartimento di PatologiaClinica ASMN


Systematic reviews• “…in science…two processes are thus atwork side by side, the reception of newmaterial and the digestion and assimilationof the old…”• Lord RayleighProfessor of Physics at CambridgeUniversity 1884baricchir Dipartimento di PatologiaClinica ASMN


Summary point• Narrative Reviews are tools for health careworkers,researchers,and policy maker whowant to keep up with the evidence that isaccumulating in their fields but…• Systematic reviews allow for a moreobjective appraisal of the evidence thantraditional narrative reviews• Meta-analysis ,if appropriate,will enhancethe precision of estimates of treatmenteffects,leading to reduced probability offalse negative resultsbaricchir Dipartimento di PatologiaClinica ASMN


What’s the right name• Narrative Reviews• Systematic Reviews. : reviews that has beenprepared using a systematic approach to minimisingbiases and random errors which is documented in amaterial and methods section. With meta-analysis QuantitativeSystematic Reviews .Without meta-analysis Qualitative Systematic reviews• M.A. :a statistical technique for combining theresults from independent studies which aims toproduce a single estimate of a treatment effectbaricchir Dipartimento di PatologiaClinica ASMN


Meta-analysisanalysis• Is most often used to assess the clinicaleffectiveness of healthcare interventions bycombining data from two to morerandomized control trials• It provide a precise estimate of treatmenteffect giving due weight to the size of thedifferent studies includedbaricchir Dipartimento di PatologiaClinica ASMN


Validity of Meta-analysisanalysis• Depends on the quality of the systematicreview on which it is based• Good meta-analysis give complete coverageof all relevant studies,look for the presenceof heterogeneity and explore the robustnessof the main findings using sensitivityanalysisbaricchir Dipartimento di PatologiaClinica ASMN


aricchir Dipartimento di PatologiaClinica ASMN


aricchir Dipartimento di PatologiaClinica ASMN


Limitations of narrative reviews• Is subjective• no clear the sources of informations• no standardized methodological quality ofstudiesbaricchir Dipartimento di PatologiaClinica ASMN


Distinction in meta-analysisanalysis• A clear distinction should be made betweenmeta-analysis of RCT and meta-analysis ofepidemiological studiesbaricchir Dipartimento di PatologiaClinica ASMN


Limitations of a single study• Often fails to detect ,or exclude, adifference in the effects of two therapies• the number of patients included in trials isoften inadeguate• expl. drug < 10% mortality in IMA ; inorder to detect such an effect with 90%certainty over 10000 in each treatmentgroup are neededbaricchir Dipartimento di PatologiaClinica ASMN


Principles of systematic reviewssystematic reviews• Carefully planned with a written protocol• Formulation of a review question• A priori definition of eligibility criteria fortrials• A comprehensive search of such trials andan assessment of their methodologicalquality• There are different statistical methods forcombining the data but there is not a single“correct” methodbaricchir Dipartimento di PatologiaClinica ASMN


Principles of systematic reviews• When interpreting results the reviewersshould consider the importance of beneficaland harmful effects and address economicimplications and implications for futureresearchbaricchir Dipartimento di PatologiaClinica ASMN


Formulate review question• Define inclusion and exclusion criteria• This require a clear statement of the intervention ofinterest• Partecipants• interventions and comparisons• outcomes• study desing and methodological qualitybaricchir Dipartimento di PatologiaClinica ASMN


Locate studies• Develop a search strategy considering thefollowing sources• CCTR• electronic databeses not covered by CCTR• handsearching of key journals• personal communication with experts in the filedbaricchir Dipartimento di PatologiaClinica ASMN


Select studies• Have eligibility checked by more than oneobserver• Develop strategy to resolve disagreementbaricchir Dipartimento di PatologiaClinica ASMN


Assess study quality• Consider assessment by more than oneobserver• Use simple score• Always assess concealment of treatmentallocation, blinding• Consider blinding of observers to authors,journals and institutionsbaricchir Dipartimento di PatologiaClinica ASMN


Again about quality• Same question ?• Same trials ?• Same quality ?baricchir Dipartimento di PatologiaClinica ASMN


S.R. are the best of the best ?• What is better for health care workers:meta-analyses of small trials or a largerandomised controlled trial ?baricchir Dipartimento di PatologiaClinica ASMN


Problems may arise• A sistematic review may be done badly ; why ?• -inappropriate aggregation of studies that differ in terms ofintervention used or patients included• -the findings from s.r. are not always in harmony with thefindings from large scale high quality single trialbaricchir Dipartimento di PatologiaClinica ASMN


Extract databaricchir Dipartimento di PatologiaClinica ASMN


Analyse and present resultsbaricchir Dipartimento di PatologiaClinica ASMN


Analyse and present resultsbaricchir Dipartimento di PatologiaClinica ASMN


Interpret results• Consider limitations and related bias• Consider strenght of evidence• Consider applicability• Consider NNT and benefit/arm• Consider economic implication• Consider implications for future researchbaricchir Dipartimento di PatologiaClinica ASMN


Type of reporting bias• Publication bias• Time lag bias• Multiple publication bias• Citation bias• Language bias• Outcome reporting biasbaricchir Dipartimento di PatologiaClinica ASMN


Identifying randomized trials• In 1993 before C.C. only 19000 reports ofC.T. were identifiable in MEDLINE• At the end of 1999 the Cochrane ControlledTrials register identify more than 250000reports of controlled trialsbaricchir Dipartimento di PatologiaClinica ASMN


Sources to be searched to identifyrandomized trials for systematicreviews• The Cochrane Controlled Trials Register• MEDLINE and EMBASE• Other database• Journals• Conference proceedings• Reference listsbaricchir Dipartimento di PatologiaClinica ASMN


Quality• design• conduct• clinical relevance• quality of reporting• Is difficult to definebaricchir Dipartimento di PatologiaClinica ASMN


Assessing the quality of randomizedcontrolled trials• Inadequate quality of studies may distort theresults from systematic reviews and metaanalyses,evaluation of studies is routinelyevaluated; this is best done using sensitivityanalysis• the use of summary score is problematic so:• -concealment of treatment allocation• -blinding• -outcome assessment• -handling of patient attritionbaricchir Dipartimento di PatologiaClinica ASMN


• Internal validityValidity of a trial• External validitybaricchir Dipartimento di PatologiaClinica ASMN


Bias• Any process at any stage of interferencetending to produce results that differsystematically from the true valuesbaricchir Dipartimento di PatologiaClinica ASMN


Randomisation consists in two parts• Generation of allocation sequences• Adequate• Inadequate• Concealment of allocation sequences• Adequate• Inadequatebaricchir Dipartimento di PatologiaClinica ASMN


Generation of allocation sequences• Adequate if resulting sequences areunpredictable• computer generated random numbers• table of random numbers• coin tossing• throwing dice• Inadequate if resulting sequences arepredictable• according to date of birth• according to date of admissionbaricchir Dipartimento di PatologiaClinica ASMN


Concealment of allocation sequences• Adequate if patients and enrollinginvestigators cannot foresee assignment• central randomisation• a priori numbered or coded containers• Inadequate if patients and investigators canforesee upcoming assignmentopen allocation scheduleall procedures based on inadequategeneration of allocation sequencesbaricchir Dipartimento di PatologiaClinica ASMN


Internal validity• Is clearly a prerequisite for externalvalidity; is the extent to which the results ofa study are correct for the circumstancesbeing studied; the aim is the reduction ofsystematic bias• selection bias• performance bias• detection bias• attrition biasbaricchir Dipartimento di PatologiaClinica ASMN


Issues addressed by Jadad’s qualityassessment scale• Randomisation• described as randommized ?• allocation sequences appropriately generated ?• Blinding• described as double blind ?• control treatment described as hidden ?• Patients attrition• described for each group(including the• number of patients lost or excluded) along• with the reasons ?baricchir Dipartimento di PatologiaClinica ASMN


External validity• The extent to which the results of a trialprovide a correct basis for applicability toother circumstances• Patients :age , sex , severity of disease , co-morbidity• Treatment regimens: dosage,timing,concomitant therapies• Settings:level of care ,experience and specialisation of careprovider• Modality of outcomes:definition of outcomes,lenght offollow upbaricchir Dipartimento di PatologiaClinica ASMN


CONSORT• Initiative to improve thequality of reportingof randomized controlled trialsbaricchir Dipartimento di PatologiaClinica ASMN


QUOROM•Quality•Of•Reporting•Of•Meta-analysees• is the “gold standard “ for reporting of S.R.baricchir Dipartimento di PatologiaClinica ASMN


QUOROM• Checklist of 18 items• flow diagrambaricchir Dipartimento di PatologiaClinica ASMN


aricchir Dipartimento di PatologiaClinica ASMN


Systematic reviews of observationalstudies• Are as common as reviews of randomizedcontrolled trials• Confounding and selection bias often distort thefindings of observational studies. Bigger is notnecessarily better• Meta-analyses of observational data can produceprecise but spurious results so statisticalcombination is not a prominent component ;moreis gained by examining possible sources ofheterogeneity between the results from ob. studiesbaricchir Dipartimento di PatologiaClinica ASMN


Confounding and bias• MRFIT• ISIS-3baricchir Dipartimento di PatologiaClinica ASMN


S.R. of prognostic variables• Prognostic studies are difficult to search• Prognostic variables should be evaluated ina sample ,of patients at a common point inthe course of their disease• Evaluation of study quality is essential• A high proportion of prognostic studies areof poor methodologic qualitybaricchir Dipartimento di PatologiaClinica ASMN


Problems with S.R. ofprognostic studies• Difficulty of identifying all studies• Inadequate reporting of methods• Variation in study design• Most studies are retrospective• Variation in methods of analysis• Lack of recognised criteria for qualityassessmentbaricchir Dipartimento di PatologiaClinica ASMN


Requirement for meta-analisysanalisys• The main requirement for a worthwhilemata analisys is first and foremost a wellexecutedsystematic reviewbaricchir Dipartimento di PatologiaClinica ASMN


Interpreting the main findings• Blobbogram• Blob or square(the measured effect)• Horizontal line(usually 95% confidence interval)• The size of blob may vary to reflect the amount of information in thatindividual study• The lenght of horizontal line is the estimate of the treatment effect forthat study• Odds ratio (summary measure)• For practical purpose is similar to relative riskbaricchir Dipartimento di PatologiaClinica ASMN


Bias in meta-analysisanalysis• Location and selection of studies• “ a statistical analysis which combinesor integrates the results of severalindependent clinical trials considered bythe analyst to be combinable”• The key difficulty lies in deciding whichset of studies are “combinable”,so goodmeta-analysis will use explicit andobjective criteria for excvlusionorrejection of studiesbaricchir Dipartimento di PatologiaClinica ASMN


Heterogeneity• Patient group• Interventions applied• Primary outcomes• Different settingsbaricchir Dipartimento di PatologiaClinica ASMN


Loss of information in importantoutcomes• Dichotomous outcomes• Pain/no pain• Dead/alive• Ecc.• Expl.• use of 50% reduction of pain• divide patients in “success” and “failure”baricchir Dipartimento di PatologiaClinica ASMN


Conflict with new experimental data• Meta-analysis vs. mega trialsbaricchir Dipartimento di PatologiaClinica ASMN


G.I.G.O. principlebaricchir Dipartimento di PatologiaClinica ASMN

More magazines by this user
Similar magazines