Evidence-Based Evidence-Based Medicine and Critical Medicine and Critical Appraisal Appraisal Ben Rehman Ben Rehman Director Director London Medicines Information London Medicines Information Service Service Northwick Park Hospital Northwick Park Hospital
Evidence-Based Medicine and Critical Appraisal. Ben Rehman Director London Medicines Information Service Northwick Park Hospital. Introduction. Evidence-Based Medicine What is it? Why do we need it? How do we do it? Critical appraisal—RCTs - PowerPoint PPT Presentation
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Evidence-Based Medicine and Evidence-Based Medicine and Critical AppraisalCritical Appraisal
Ben RehmanBen RehmanDirectorDirector
London Medicines Information ServiceLondon Medicines Information ServiceNorthwick Park HospitalNorthwick Park Hospital
IntroductionIntroduction Evidence-Based MedicineEvidence-Based Medicine
• What is it? What is it? • Why do we need it?Why do we need it?• How do we do it?How do we do it?
Critical appraisal—RCTsCritical appraisal—RCTs Brief introduction to other study designs including Brief introduction to other study designs including
types of pharmacoeconomic studiestypes of pharmacoeconomic studies
• How should we use it?How should we use it? At patient level?At patient level? As part of broader decision making and resource As part of broader decision making and resource
optimisation? optimisation?
Example critical appraisal workshopExample critical appraisal workshop
Evidence-Based Medicine—Evidence-Based Medicine—What is it?What is it?
the conscientious, systematic, explicit, and the conscientious, systematic, explicit, and judicious use of current best evidence in judicious use of current best evidence in patient carepatient care
best research evidence replaces or best research evidence replaces or supports current practice supports current practice
integration of evidence and patient or integration of evidence and patient or population values population values
optimising health gainoptimising health gain• clinical evidenceclinical evidence• pharmacoeconomicspharmacoeconomics
Evidence-Based Medicine—Evidence-Based Medicine—Why do we need it?Why do we need it?
Health professionals face problems Health professionals face problems ensuring best possible patient care:ensuring best possible patient care:• information overloadinformation overload• inadequate traditional sourcesinadequate traditional sources• opinion based approach flawedopinion based approach flawed• disparity between experience and up-to-date disparity between experience and up-to-date
knowledgeknowledge Healthcare systems have finite resourcesHealthcare systems have finite resources But… it’s not a complete panacea!But… it’s not a complete panacea!
Evidence-Based Medicine—Evidence-Based Medicine—How do we do it?How do we do it?
1.1. Form an answerable questionForm an answerable question
2.2. Obtain current best evidenceObtain current best evidence
3.3. Critically appraise the evidenceCritically appraise the evidence
4.4. Integrate appraised evidence with Integrate appraised evidence with clinical expertise and patient valuesclinical expertise and patient values
5.5. Evaluate our successes and failuresEvaluate our successes and failures
Evidence-Based Medicine—Evidence-Based Medicine—How do we do it?How do we do it?
Formulate question
Integrate evidence and values—implement
changes
Evaluate successes and failures—audit
Track down best evidence
Critically review quality
How do we do EBM?How do we do EBM?
1.1. Form an answerable questionForm an answerable question Should contain:Should contain:
• Intervention you are interested inIntervention you are interested in• Patient or population you are interested Patient or population you are interested
in (clearly for drug interventions you in (clearly for drug interventions you require the indication here too)require the indication here too)
• Outcomes you are interested inOutcomes you are interested in So for the example paper, our So for the example paper, our
original question might have been:original question might have been:• Is agomelatine effective in treating Is agomelatine effective in treating
major depressive disorder?major depressive disorder?
How do we do EBM?How do we do EBM?
2.2. Obtain current best evidenceObtain current best evidence Group discussionGroup discussion
• What are good sources of evidence?What are good sources of evidence? What have you used in the past and found What have you used in the past and found
useful?useful?
• What makes them good?What makes them good? What criteria do you apply?What criteria do you apply?
• How can we find good evidence?How can we find good evidence? What portals and sources do we have What portals and sources do we have
available?available?
How do we do EBM?How do we do EBM?
2.2. Obtain current best evidenceObtain current best evidence Discussion point: good sources of Discussion point: good sources of
evidence for EBMevidence for EBM
How do we do EBM?How do we do EBM?
2.2. Obtain current best evidenceObtain current best evidence An appropriate search strategy is An appropriate search strategy is
importantimportant We should be aware of the hierarchy We should be aware of the hierarchy
Are this study’s results valid?Are this study’s results valid?• Did the trial address a clearly focused issue?Did the trial address a clearly focused issue?
Consider the intervention and the outcomes Consider the intervention and the outcomes consideredconsidered
• Is an RCT an appropriate method to answer Is an RCT an appropriate method to answer this question?this question?
• Was the assignment of patients to treatment Was the assignment of patients to treatment randomised?randomised?
Why is it important?Why is it important? Acceptable vs dubious methods?Acceptable vs dubious methods? Is the process well described?Is the process well described? Was the process concealed?Was the process concealed?
Are this study’s results valid?Are this study’s results valid?• Were the groups similar at the start of the Were the groups similar at the start of the
trial?trial? Could any differences have affected outcome?Could any differences have affected outcome? Was any method used to balance randomisation Was any method used to balance randomisation
(stratification)?(stratification)?• Was follow-up sufficient?Was follow-up sufficient?
LengthLength CompletenessCompleteness
• Was there sufficient power?Was there sufficient power? Were there enough participants to minimise the play Were there enough participants to minimise the play
of chance?of chance?• Was an intention-to-treat analysis performed?Was an intention-to-treat analysis performed?
Are the results of this individual study Are the results of this individual study important for us?important for us?• Define the population studiedDefine the population studied
Inclusion and exclusion criteriaInclusion and exclusion criteria
• EndpointsEndpoints Measurement of outcomeMeasurement of outcome
• Clinical relevanceClinical relevance• Surrogate and composite markersSurrogate and composite markers• Validity Validity
Primary vs. secondaryPrimary vs. secondary
• Balance of beneficial outcomes against side-Balance of beneficial outcomes against side-effectseffects
Are the results of this individual study Are the results of this individual study important for us?important for us?• What is the magnitude of the treatment effect?What is the magnitude of the treatment effect?
How is it described?How is it described?• Proportions of people, a measurement (mean or Proportions of people, a measurement (mean or
median differences), a survival curve?median differences), a survival curve? How is it expressed?How is it expressed?
• With proportions see terms like relative risk reduction, With proportions see terms like relative risk reduction, absolute risk reduction, number needed to treatabsolute risk reduction, number needed to treat
• Is what is being expressed clear?Is what is being expressed clear?• Is it clinically significant?Is it clinically significant?
Definitions for measures of Definitions for measures of association and effectivenessassociation and effectiveness• Control event rateControl event rate
the rate at which an outcome occurs in the control the rate at which an outcome occurs in the control populationpopulation
• Experimental event rateExperimental event rate the rate at which an outcome occurs in the the rate at which an outcome occurs in the
experimental populationexperimental population
Definitions for measures of association Definitions for measures of association and effectivenessand effectiveness• Absolute risk (AR)Absolute risk (AR)
Probability an individual will experience a specified Probability an individual will experience a specified outcome during a specified periodoutcome during a specified period
Range of 0 to 1, or expressed as a percentageRange of 0 to 1, or expressed as a percentage
• Relative risk (RR)Relative risk (RR) How many times more likely (RR > 1) or less likely (RR < 1) How many times more likely (RR > 1) or less likely (RR < 1)
an event is to happen in one group compared with another an event is to happen in one group compared with another
• Absolute risk reduction (ARR)Absolute risk reduction (ARR) Absolute difference in risk between experimental and Absolute difference in risk between experimental and
control groups control groups
• Relative risk reduction (RRR)Relative risk reduction (RRR) Proportional reduction in risk between experimental and Proportional reduction in risk between experimental and
control participants in a trial control participants in a trial
Definitions for measures of Definitions for measures of association and effectivenessassociation and effectiveness• Number needed to treat (NNT)Number needed to treat (NNT)
A measure of treatment effectiveness A measure of treatment effectiveness Measures the people who need to be treated with an Measures the people who need to be treated with an
intervention over a period of time to prevent an intervention over a period of time to prevent an additional adverse outcome or achieve an additional additional adverse outcome or achieve an additional beneficial outcome beneficial outcome
Definitions for measures of association Definitions for measures of association and effectivenessand effectiveness• Odds ratio (OR)Odds ratio (OR)
Measure of treatment effectiveness—likelihood of event Measure of treatment effectiveness—likelihood of event happening in experimental group vs. control grouphappening in experimental group vs. control group
Effects being measured may be adverse (e.g. death or Effects being measured may be adverse (e.g. death or disability) or desirable (e.g. survival) disability) or desirable (e.g. survival)
If events rare OR analogous to relative risk (RR); as If events rare OR analogous to relative risk (RR); as event rates increase the OR and RR divergeevent rates increase the OR and RR diverge
• When interpreting remember:When interpreting remember: Closer OR is to 1 Closer OR is to 1 smaller the difference in effect smaller the difference in effect
between experimental intervention and control between experimental intervention and control interventionintervention
OR > 1 ≡ effects of treatment more than controlOR > 1 ≡ effects of treatment more than control OR < 1 ≡ effects of treatment less than control OR < 1 ≡ effects of treatment less than control OR 95% CI includes 1 ≡ effect cannot be deemed OR 95% CI includes 1 ≡ effect cannot be deemed
Absolute vs. relative risk reduction as Absolute vs. relative risk reduction as measures—an examplemeasures—an example• Pros and cons of each measurePros and cons of each measure
Clinical significance?Clinical significance? Proportional difference?Proportional difference? Ease of interpretation?Ease of interpretation?
An example—4S study (Scandinavian An example—4S study (Scandinavian Simvastatin Survival Study)Simvastatin Survival Study)11 • PurposePurpose
Evaluate effect of cholesterol lowering with simvastatin Evaluate effect of cholesterol lowering with simvastatin on mortality and morbidity in patients with coronary on mortality and morbidity in patients with coronary heart disease (CHD) heart disease (CHD)
• ParticipantsParticipantsPatients with stable angina or history of myocardial Patients with stable angina or history of myocardial infarction more than 6 months previouslyinfarction more than 6 months previouslySerum cholesterol > 6.2mmol/lSerum cholesterol > 6.2mmol/lExcluded patients with arrhythmias and heart failureExcluded patients with arrhythmias and heart failureRun in of 8 weeks of dietary therapyRun in of 8 weeks of dietary therapyIf cholesterol still raised (5.5-8.0 mmol/L) randomised to If cholesterol still raised (5.5-8.0 mmol/L) randomised to receive simvastatin (20mg esculating to 40mg) or receive simvastatin (20mg esculating to 40mg) or placebo placebo
1.1. Lancet. 1994 Nov 19;344(8934):1383-9Lancet. 1994 Nov 19;344(8934):1383-9
4S study4S study• Randomisation and allocationRandomisation and allocation
4444 patients randomised to double-blind 4444 patients randomised to double-blind treatment with simvastatin or placebotreatment with simvastatin or placebo
2223 patients placebo 2223 patients placebo
2221 were simvastatin2221 were simvastatin
Mean follow-up was 5.4 yearsMean follow-up was 5.4 years
622 patients (28%) in placebo group vs. 431 (19%) in 622 patients (28%) in placebo group vs. 431 (19%) in the simvastatin group the simvastatin group
• So, on the face of it, the results look extremely So, on the face of it, the results look extremely promising. How do the measures of association we promising. How do the measures of association we discussed earlier play out?discussed earlier play out?
NNT examples—when should we adopt therapy?NNT examples—when should we adopt therapy?
Streptokinase + aspririn v. placebo (ISIS 2)
prevent 1 death at 5 weeks
20
tPA v. streptokinase (GUSTO trial)
save 1 life with tPA usage
100
Simvastatin v. placebo in IHD (4S study)
prevent 1 event in 5y
15
Treating hypertension in the over-60s
prevent 1 event in 5y
18
Aspirin v. placebo in healthy adults
prevent MI or death in 1 year
500
InterventionIntervention OutcomeOutcome NNNNTT
Discussion point: numbers needed to Discussion point: numbers needed to treattreat• When do you think therapy should be When do you think therapy should be
adopted? adopted? • Clearly there is no one answer to this Clearly there is no one answer to this
but what might be the considerations?but what might be the considerations?
Statistical approaches to uncertaintyStatistical approaches to uncertainty• Why do they exist?Why do they exist?
Cannot include all individuals in a population Cannot include all individuals in a population in trialin trial
Need to quantify uncertaintyNeed to quantify uncertainty
• p values and confidence intervals are p values and confidence intervals are the measures used, but what are they the measures used, but what are they and what are the differences?and what are the differences?
Statistical approaches to uncertaintyStatistical approaches to uncertainty• p valuesp values
assess the significance of the difference between assess the significance of the difference between a sample estimate and a hypothesised valuea sample estimate and a hypothesised value
tell us the probability that an observed effect tell us the probability that an observed effect occurred by chance if in truth there is no effectoccurred by chance if in truth there is no effect
But… doesn’t quantify the size of an effect, nor But… doesn’t quantify the size of an effect, nor the directionthe direction
p<0.05 p<0.05 → commonly reject null hypothesis→ commonly reject null hypothesis ideally p<<<0.05 and trial reports the actual ideally p<<<0.05 and trial reports the actual
Statistical approaches to uncertaintyStatistical approaches to uncertainty• Confidence intervalsConfidence intervals
Range of values around the sample estimate Range of values around the sample estimate (i.e. the value found in the study)(i.e. the value found in the study)
The range has specified probability (usually The range has specified probability (usually 95%) so CI acts as hypothesis test for the 95%) so CI acts as hypothesis test for the rangerange
Should be seen with most measures of Should be seen with most measures of associationassociation
Statistical approaches to uncertainty—Statistical approaches to uncertainty—please remember!please remember!• statistical significance statistical significance ≠≠ clinical significance… clinical significance…
what is actually being measured?what is actually being measured?• sample size is still importantsample size is still important
larger sample = less uncertainty = narrower CI larger sample = less uncertainty = narrower CI and smaller observed effect considered significantand smaller observed effect considered significant
• there can still be errorthere can still be error rejecting null hypothesis when it’s true (type 1)rejecting null hypothesis when it’s true (type 1) Not rejecting the null hypothesis when it’s false Not rejecting the null hypothesis when it’s false
Increasingly used by pharma. Increasingly used by pharma. Designed to show a new treatment is Designed to show a new treatment is
not inferior to standard treatment by not inferior to standard treatment by a predefined clinically acceptable a predefined clinically acceptable endpointendpoint
Rely on assumptions that can be Rely on assumptions that can be hard to validatehard to validate
How do we do EBM?How do we do EBM?
3.3. Critical appraisal—RCTs of Critical appraisal—RCTs of therapeutic equivalencetherapeutic equivalence
Trials based on the concept that:Trials based on the concept that:• new treatment is non-inferiornew treatment is non-inferior• but would exhibit therapeutic efficacy if but would exhibit therapeutic efficacy if
a placebo controlled RCT could/was a placebo controlled RCT could/was performedperformed
• new treatment offers ancillary benefitsnew treatment offers ancillary benefits Note: non-inferior so efficacy Note: non-inferior so efficacy
determined by effect against placebo determined by effect against placebo not comparatornot comparator
How do we do EBM?How do we do EBM?
3.3. Critical appraisal—RCTs of Critical appraisal—RCTs of therapeutic equivalencetherapeutic equivalence
Estimation of non-inferiority marginEstimation of non-inferiority margin• Use statistical and clinical reasoning to Use statistical and clinical reasoning to
determine what is non-inferiordetermine what is non-inferior• If there are a variety of placebo If there are a variety of placebo
controlled RCTs this will be similar to a controlled RCTs this will be similar to a meta-analysis meta-analysis
Non-inferiority margin determined as Non-inferiority margin determined as a fraction (f) and specifies an a fraction (f) and specifies an acceptable magnitude for treatment acceptable magnitude for treatment effect that must be preservedeffect that must be preserved
How do we do EBM?How do we do EBM?
3.3. Critical appraisal—RCTs of Critical appraisal—RCTs of therapeutic equivalencetherapeutic equivalence
What to look forWhat to look for• These trials should be designed in a manner These trials should be designed in a manner
consistent with the historical placebo-controlled consistent with the historical placebo-controlled trialstrials
• The active comparator should be well established The active comparator should be well established with predictable quantifiable, and consistent effectswith predictable quantifiable, and consistent effects
• What constitutes non-inferiority should be What constitutes non-inferiority should be determined prior to initiation of trialdetermined prior to initiation of trial
• Protocol deviations and poor adherence may have a Protocol deviations and poor adherence may have a larger impact on quality than in conventional trialslarger impact on quality than in conventional trials
• Analysis should be by intention-to-treat AND on-Analysis should be by intention-to-treat AND on-treatment, and should also report absolute AND treatment, and should also report absolute AND relative effects relative effects
How do we do EBM?How do we do EBM?
3.3. Critical appraisal—RCTs of Critical appraisal—RCTs of therapeutic equivalencetherapeutic equivalence
So when appraising these trials ask So when appraising these trials ask yourself:yourself:• Was the active control previously shown to be effective?Was the active control previously shown to be effective?• Were study patients and outcome variables similar to Were study patients and outcome variables similar to
those in the original trials that established the efficacy of those in the original trials that established the efficacy of the active control?the active control?
• Were both regimens applied in optimal fashion?Were both regimens applied in optimal fashion?• Was the appropriate null hypothesis tested?Was the appropriate null hypothesis tested?• Was the equivalence margin specified before the study?Was the equivalence margin specified before the study?• Was the trial large enough?Was the trial large enough?• Was the analysis intention-to-treat AND on-treatment, Was the analysis intention-to-treat AND on-treatment,
and should also report absolute AND relative effects?and should also report absolute AND relative effects?• PLUS usual assessment of size and precision of PLUS usual assessment of size and precision of
treatment effect!treatment effect!
How do we do EBM?How do we do EBM?
3.3. Critical appraisal—RCTs of Critical appraisal—RCTs of therapeutic equivalencetherapeutic equivalence
Many other types of study in the Many other types of study in the hierarchy e.g. systematic reviews, hierarchy e.g. systematic reviews, case-control studies, cohort studiescase-control studies, cohort studies
Details of appraisal not included Details of appraisal not included today but worth thinking about most today but worth thinking about most in terms of:in terms of:• Internal validityInternal validity• BiasBias• External validityExternal validity
How do we do EBM?How do we do EBM?
3.3. Critical appraisal—other Critical appraisal—other study designsstudy designs
How do we do EBM?How do we do EBM?
3.3. Critical appraisal—other Critical appraisal—other study designsstudy designs
Cohort studyCohort study
PopulationSample
Exposed
Not exposed
Outcome
Outcome
Time
Case control studyCase control study
How do we do EBM?How do we do EBM?
3.3. Critical appraisal—other Critical appraisal—other study designsstudy designs
Population
CasesExposed
Not exposed
Time
ControlsNot exposed
Exposed
Study
Important in decision makingImportant in decision making Considers costs and consequences of Considers costs and consequences of
alternate courses of actionalternate courses of action Scope depends on level of decision Scope depends on level of decision
making they informmaking they inform Costs always measured the same Costs always measured the same
way (although scope may vary), way (although scope may vary), measurement of consequence varies measurement of consequence varies according to study typeaccording to study type
How do we do EBM?How do we do EBM?
3.3. Economic evidenceEconomic evidence
Cost minimisation analysesCost minimisation analyses• analyse difference in costs where there analyse difference in costs where there
is no difference in outcomeis no difference in outcome• narrow focusnarrow focus
Cost-effectiveness analysesCost-effectiveness analyses• outcome expressed in natural units (e.g. outcome expressed in natural units (e.g.
validated single marker)validated single marker)• again fairly narrow scope—generally 2 again fairly narrow scope—generally 2
interventions for single indicationinterventions for single indication
How do we do EBM?How do we do EBM?
3.3. Critical appraisal—what is Critical appraisal—what is economic evidence?economic evidence?
Cost-utility analysesCost-utility analyses• Costs and consequences but Costs and consequences but
consequence measured as utility consequence measured as utility • Utility = a value of health state rather Utility = a value of health state rather
than a natural marker (usually QALY)than a natural marker (usually QALY)• QALY is a measure of a person’s length QALY is a measure of a person’s length
of life weighted by a valuation of their of life weighted by a valuation of their health-related quality-of-life (HRQL) over health-related quality-of-life (HRQL) over that periodthat period
• Can compare incongruent interventionsCan compare incongruent interventions
How do we do EBM?How do we do EBM?
3.3. Critical appraisal—what is Critical appraisal—what is economic evidence?economic evidence?
Cost utility analysisCost utility analysis• Used by NICEUsed by NICE
Consider outcome in terms valuable to patients i.e. Consider outcome in terms valuable to patients i.e. life expectancy and qualitylife expectancy and quality
Uses a standard measure (utility)Uses a standard measure (utility) Comprehensive consideration of costsComprehensive consideration of costs NICE state that:NICE state that:
Technologies can be considered to be cost Technologies can be considered to be cost effective if their health benefits are greater effective if their health benefits are greater than the opportunity costs measured in terms than the opportunity costs measured in terms of the health benefits associated with of the health benefits associated with programmes that may be displaced to fund the programmes that may be displaced to fund the new technology. In other words, the general new technology. In other words, the general consequences for the wider group of patients consequences for the wider group of patients in the NHS are considered alongside the effects in the NHS are considered alongside the effects for those patients who may directly benefit for those patients who may directly benefit from the technology of interest.from the technology of interest.
How do we do EBM?How do we do EBM?
3.3. Critical appraisal—what is Critical appraisal—what is economic evidence?economic evidence?
Cost-benefit analysesCost-benefit analyses• Broadest possible scope!Broadest possible scope!• Allocation of resources between Allocation of resources between
difference sectors of economydifference sectors of economy• Maximising social welfareMaximising social welfare• Rely on assigning cost to healthcare Rely on assigning cost to healthcare
intervention—incredible complicated in intervention—incredible complicated in practice, particular in NHS where practice, particular in NHS where individuals don’t pay at point of deliveryindividuals don’t pay at point of delivery
How do we do EBM?How do we do EBM?
3.3. Critical appraisal—what is Critical appraisal—what is economic evidence?economic evidence?
A lot is published! But texts I’ve found A lot is published! But texts I’ve found useful include:useful include:• Straus SE, Richardson WS, Paul Glasziou, Haynes RB. Straus SE, Richardson WS, Paul Glasziou, Haynes RB.
Evidence-based Medicine: How to Practice and Teach Evidence-based Medicine: How to Practice and Teach EBM, Third Edition. Churchill Livingstone: Edinburgh, EBM, Third Edition. Churchill Livingstone: Edinburgh, 20052005
• Guyatt GH, et al. Users’ Guides to the Medical Literature: Guyatt GH, et al. Users’ Guides to the Medical Literature: II. How to use an article about therapy or prevention. A. II. How to use an article about therapy or prevention. A. Are the results of the study valid? JAMA 1993; 270: Are the results of the study valid? JAMA 1993; 270: 2598-26012598-2601
• Guyatt GH, et al. Users’ Guides to the Medical Literature: Guyatt GH, et al. Users’ Guides to the Medical Literature: II. How to use an article about therapy or prevention. B. II. How to use an article about therapy or prevention. B. What are the results and will they help me in caring for What are the results and will they help me in caring for my patients? JAMA 1994; 271: 59-63my patients? JAMA 1994; 271: 59-63
• Evidence Based Medicine (EBM) journal notebook seriesEvidence Based Medicine (EBM) journal notebook series—available online —available online
Is our patient/population so different from Is our patient/population so different from those in the study that its results cannot those in the study that its results cannot apply?apply?
Is the treatment feasible for us?Is the treatment feasible for us? What are the potential benefits and harms What are the potential benefits and harms
from therapy?from therapy? What are our patients’ values and What are our patients’ values and
preferences for both the outcome and preferences for both the outcome and side-effects?side-effects?
Is there any economic analysis supporting Is there any economic analysis supporting therapy?therapy?
How do we use EBM?How do we use EBM?
4.4. Applying results of research to Applying results of research to our individual patient or populationour individual patient or population
Example critical appraisal of RCT Example critical appraisal of RCT workshopworkshop
The CASP toolThe CASP tool• available at available at
• Divide up into groups and use the tool to Divide up into groups and use the tool to appraise the trial:appraise the trial:
Placebo-controlled trial of agomelatine Placebo-controlled trial of agomelatine in the treatment of major depressive in the treatment of major depressive disorderdisorder
Example critical appraisal of RCT Example critical appraisal of RCT workshopworkshop
Feedback from groups about the trialFeedback from groups about the trial